Remove legacy aliases and backward-compat shims from config schema

- Remove dead schema fields: rotate90_axes, val_split, train_resolution,
  max_angle, pad_size (SlidingWindowConfig), val_check_interval (ScalarLoggingConfig)
- Remove compute_bbox_all_2d/3d aliases (only compute_bbox_all is used)
- Fix misleading "legacy" comments on active fields (keep_input_on_cpu,
  rotation90_axes, MotionBlurConfig, val_check_interval on optimization)
- Change rotation90_axes type to Any to support "all" string value
- Auto-generate SQLite storage for Optuna studies when save_study=True
- Update 13 tutorial YAMLs to use canonical field names
- Includes linter auto-formatting across the codebase

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: Donglai Wei

connectomics/config/__init__.py

@@ -2,41 +2,40 @@
 Modern Hydra-based configuration system for PyTorch Connectomics.
 """
 
-# New Hydra config system (primary)
-from .schema import Config
-from .pipeline.config_io import (
-    load_config,
-    save_config,
-    merge_configs,
-    update_from_cli,
-    to_dict,
-    from_dict,
-    print_config,
-    validate_config,
-    get_config_hash,
-    create_experiment_name,
-    resolve_data_paths,
-)
-from .pipeline.stage_resolver import resolve_default_profiles
-from .pipeline.dict_utils import to_plain, as_plain_dict, cfg_get
-
 # Auto-configuration system
 from .hardware.auto_config import (
-    auto_plan_config,
     AutoConfigPlanner,
     AutoPlanResult,
+    auto_plan_config,
     resolve_runtime_resource_sentinels,
 )
 
 # GPU utilities
 from .hardware.gpu_utils import (
+    estimate_gpu_memory_required,
     get_gpu_info,
+    get_optimal_num_workers,
     print_gpu_info,
     suggest_batch_size,
-    estimate_gpu_memory_required,
-    get_optimal_num_workers,
 )
+from .pipeline.config_io import (
+    create_experiment_name,
+    from_dict,
+    get_config_hash,
+    load_config,
+    merge_configs,
+    print_config,
+    resolve_data_paths,
+    save_config,
+    to_dict,
+    update_from_cli,
+    validate_config,
+)
+from .pipeline.dict_utils import as_plain_dict, cfg_get, to_plain
+from .pipeline.stage_resolver import resolve_default_profiles
 
+# New Hydra config system (primary)
+from .schema import Config
 
 __all__ = [
     # Hydra config system

connectomics/config/hardware/__init__.py

@@ -1,8 +1,19 @@
 """GPU-aware auto-planning and SLURM cluster utilities."""
 
-from .auto_config import auto_plan_config, AutoConfigPlanner, AutoPlanResult, resolve_runtime_resource_sentinels
-from .gpu_utils import get_gpu_info, print_gpu_info, suggest_batch_size, estimate_gpu_memory_required, get_optimal_num_workers
 from . import slurm_utils
+from .auto_config import (
+    AutoConfigPlanner,
+    AutoPlanResult,
+    auto_plan_config,
+    resolve_runtime_resource_sentinels,
+)
+from .gpu_utils import (
+    estimate_gpu_memory_required,
+    get_gpu_info,
+    get_optimal_num_workers,
+    print_gpu_info,
+    suggest_batch_size,
+)
 
 __all__ = [
     "auto_plan_config",

connectomics/config/hardware/auto_config.py

@@ -11,25 +11,26 @@
 Users can manually override any auto-determined parameters.
 """
 
-
 from __future__ import annotations
+
 import logging
-import numpy as np
-from typing import Dict, List, Optional, Any
-from dataclasses import dataclass, field
-from omegaconf import OmegaConf, DictConfig
-import warnings
 import os
+import warnings
+from dataclasses import dataclass, field
+from typing import Any, Dict, List, Optional
 
-logger = logging.getLogger(__name__)
+import numpy as np
+from omegaconf import DictConfig, OmegaConf
 
 from .gpu_utils import (
+    estimate_gpu_memory_required,
     get_gpu_info,
-    suggest_batch_size,
     get_optimal_num_workers,
-    estimate_gpu_memory_required,
+    suggest_batch_size,
 )
 
+logger = logging.getLogger(__name__)
+
 
 def _available_cpus_for_current_run() -> int:
     """
@@ -120,7 +121,9 @@ def resolve_runtime_resource_sentinels(
             logger.info(
                 "Auto-detected system.num_workers: -1 -> %d "
                 "(available_cpus=%d, local_processes=%d)",
-                config.system.num_workers, available_cpus, process_count,
+                config.system.num_workers,
+                available_cpus,
+                process_count,
             )
 
     if getattr(config.system, "num_gpus", 0) < -1:
@@ -477,7 +480,10 @@ def auto_plan_config(
     # Collect manual overrides (values explicitly set in config)
     manual_overrides = {}
     if hasattr(config, "data"):
-        if hasattr(config.data, "dataloader") and getattr(config.data.dataloader, "batch_size", None) is not None:
+        if (
+            hasattr(config.data, "dataloader")
+            and getattr(config.data.dataloader, "batch_size", None) is not None
+        ):
             manual_overrides["batch_size"] = config.data.dataloader.batch_size
         if hasattr(config, "system") and getattr(config.system, "num_workers", None) is not None:
             manual_overrides["num_workers"] = config.system.num_workers

connectomics/config/hardware/gpu_utils.py

@@ -5,13 +5,14 @@
 and estimate memory requirements for training.
 """
 
-
 from __future__ import annotations
-import torch
-import psutil
+
 import warnings
 from typing import Any, Dict, Tuple
 
+import psutil
+import torch
+
 
 def get_gpu_info() -> Dict[str, Any]:
     """

connectomics/config/hardware/slurm_utils.py

@@ -1,15 +1,15 @@
 """SLURM cluster resource detection and management utilities."""
 
-
 from __future__ import annotations
-import subprocess
-import re
+
 import json
+import logging
+import re
+import subprocess
 import time
+from dataclasses import asdict, dataclass
 from pathlib import Path
-from dataclasses import dataclass, asdict
 from typing import Dict, List, Optional
-import logging
 
 logger = logging.getLogger(__name__)
 

connectomics/config/pipeline/__init__.py

@@ -1,20 +1,20 @@
 """Core config loading, profile resolution, and stage merging."""
 
 from .config_io import (
+    create_experiment_name,
+    from_dict,
+    get_config_hash,
     load_config,
-    save_config,
     merge_configs,
-    update_from_cli,
-    to_dict,
-    from_dict,
     print_config,
-    validate_config,
-    get_config_hash,
-    create_experiment_name,
     resolve_data_paths,
+    save_config,
+    to_dict,
+    update_from_cli,
+    validate_config,
 )
+from .dict_utils import as_plain_dict, cfg_get, to_plain
 from .stage_resolver import resolve_default_profiles
-from .dict_utils import to_plain, as_plain_dict, cfg_get
 
 __all__ = [
     "load_config",

connectomics/config/pipeline/config_io.py

@@ -8,11 +8,11 @@
 
 import dataclasses
 import hashlib
-from pathlib import Path
-import warnings
 import os
 import re
+import warnings
 from glob import glob
+from pathlib import Path
 from typing import Any, Dict, List, Optional, Tuple, Union
 
 from omegaconf import DictConfig, ListConfig, OmegaConf
@@ -24,7 +24,6 @@
 from .profile_engine import _YAML_PROFILE_ENGINE
 from .stage_resolver import _collect_explicit_paths
 
-
 # ---------------------------------------------------------------------------
 # Config loading helpers
 # ---------------------------------------------------------------------------
@@ -414,7 +413,9 @@ def _validate_cross_section_coherence(cfg: Config) -> None:
                     context=f"inference.decoding[{i}].kwargs.{key}",
                 )
                 if min_channels is not None:
-                    required_channels.append((f"inference.decoding[{i}].kwargs.{key}", min_channels))
+                    required_channels.append(
+                        (f"inference.decoding[{i}].kwargs.{key}", min_channels)
+                    )
 
     # 2d) TTA channel selectors
     tta_cfg = getattr(cfg.inference, "test_time_augmentation", None)
@@ -450,15 +451,14 @@ def _validate_cross_section_coherence(cfg: Config) -> None:
         context="inference.test_time_augmentation.select_channel",
     )
     if min_channels is not None:
-        required_channels.append(
-            ("inference.test_time_augmentation.select_channel", min_channels)
-        )
+        required_channels.append(("inference.test_time_augmentation.select_channel", min_channels))
 
     if required_channels:
         required_max = max(req for _, req in required_channels)
         if required_max > out_channels:
             details = ", ".join(
-                f"{path} needs >= {req}" for path, req in sorted(required_channels, key=lambda x: x[1], reverse=True)
+                f"{path} needs >= {req}"
+                for path, req in sorted(required_channels, key=lambda x: x[1], reverse=True)
             )
             raise ValueError(
                 "Cross-section validation failed: model.out_channels is "
@@ -467,7 +467,9 @@ def _validate_cross_section_coherence(cfg: Config) -> None:
             )
 
     # 3) deep_supervision=True with architectures that don't support it
-    deep_supervision = getattr(model_loss_cfg, "deep_supervision", False) if model_loss_cfg else False
+    deep_supervision = (
+        getattr(model_loss_cfg, "deep_supervision", False) if model_loss_cfg else False
+    )
     if deep_supervision:
         arch_type = getattr(cfg.model.arch, "type", "")
         if not _architecture_supports_deep_supervision(arch_type):
@@ -561,6 +563,7 @@ def resolve_data_paths(cfg: Config) -> Config:
         >>> print(cfg.data.test.image)
         ['/data/test/volume_1.tif', '/data/test/volume_2.tif']
     """
+
     def _combine_path(
         base_path: str, file_path: Optional[Union[str, List[str]]]
     ) -> Optional[Union[str, List[str]]]: