fix: relax warmup_ratio lower bound and regenerate schema

Reviewer flagged that `gt=0` rejects the legal `warmup_ratio=0.0` config
(disable warmup). Relax to `ge=0`; `lt=1` is kept because that's the
v5 boundary where warmup_steps flips from ratio to raw step count.

Regenerate the published JSON schema so it reflects the constraint —
otherwise YAML authoring against the schema would pass schema
validation and fail at runtime.

Pushed back on the reviewer's claim that `warmup_steps=0.1` runs zero
warmup: transformers v5 typed `warmup_steps: float` and `get_warmup_steps`
branches on `>= 1`, not `> 0` — `0.1` takes the `math.ceil(N * 0.1)`
fraction branch (training_args.py:2089 in v5.12.1).

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Author: voorhs

docs/optimizer_search_space_config.schema.json

@@ -1302,6 +1302,8 @@
                 },
                 "warmup_ratio": {
                     "default": 0.1,
+                    "exclusiveMaximum": 1,
+                    "minimum": 0,
                     "title": "Warmup Ratio",
                     "type": "number"
                 },

src/autointent/configs/_transformers.py

@@ -29,9 +29,9 @@ class EmbedderFineTuningConfig(BaseModel):
     margin: float = Field(default=0.5)
     learning_rate: float = Field(default=2e-5)
     # Fed to TrainingArguments.warmup_steps in v5, which interprets float<1 as
-    # a fraction of total steps and float>=1 as a raw step count. Restrict to
-    # (0, 1) so warmup_ratio=1.0 doesn't silently become a single step.
-    warmup_ratio: float = Field(default=0.1, gt=0, lt=1)
+    # a fraction of total steps and float>=1 as a raw step count. Cap below 1
+    # so warmup_ratio=1.0 doesn't silently become a single step.
+    warmup_ratio: float = Field(default=0.1, ge=0, lt=1)
     early_stopping_patience: int = Field(default=1)
     early_stopping_threshold: float = Field(default=0.0)
     val_fraction: float = Field(default=0.2)