ruff: move per-file-ignores into inline noqa headers (#1457)

## Why

Internal coding guidelines don't allow the centralized
`[tool.ruff.lint.per-file-ignores]` block in `pyproject.toml`.
Exceptions must be annotated inline on the files that need them. Most of
the affected files are generated, so the noqa headers have to come from
the codegen templates.

## What

Removes the `[tool.ruff.lint.per-file-ignores]` block from
`pyproject.toml` and replaces it with inline `# ruff: noqa: <codes>`
headers on the affected files.

## Where the noqas come from

| File set | Codes | Source |
|---|---|---|
| `databricks/sdk/__init__.py` | F401 F403 F405 F811 | codegen template
(`pysdkv0/templates/batch/init.py.tmpl`) |
| `databricks/sdk/service/*.py` + `tests/databricks/sdk/service/*.py` |
F811 F841 | codegen template
(`pysdkv0/templates/packages/service.py.tmpl`) |
| `databricks/sdk/errors/overrides.py` | F403 F405 | codegen template
(`pysdkv0/templates/batch/error_overrides.py.tmpl`) |
| `tests/generated/test_*.py` | F403 F405 | codegen template
(`pysdkv0/unittestgen/{http,lro,marshall,idempotency}`) |
| `databricks/sdk/core.py` | F403 F405 | hand-written (star-imports for
backwards compat) |
| `databricks/sdk/errors/__init__.py` | F401 F403 F405 F811 |
hand-written (public re-export surface) |
| `databricks/sdk/logger/__init__.py` | F401 | hand-written (re-exports
`RoundTrip`) |
| `databricks/sdk/runtime/__init__.py` line 127 | F401 | hand-written
(intentional optional import probe) |

**F401 stays uncovered on `service/*.py` on purpose**: `make fmt` runs
`ruff check --fix-only` to strip the fat-import header emitted by the
codegen template. Ignoring F401 there would defeat that (the
autoflake-equivalent behavior the SDK relies on).

## Paired PR

Template changes land in databricks-eng/universe#2008679. Regenerate via
`bazel build //openapi/genkit` then `genkit update-sdk --dir .` from
this checkout. Generation job runs once a day, so the two PRs are safe
to merge outside of that window.

## Verification

```
$ make lint
uv run ruff check databricks tests
All checks passed!
uv run ruff format --check databricks tests
152 files already formatted
```

This pull request and its description were written by Isaac.

NO_CHANGELOG=true

Author: hectorcast-db

databricks/sdk/__init__.py

@@ -1,11 +1,14 @@
+# Star-import re-exports preserve backwards compatibility: config/* and
+# credentials_provider/* used to live in this module and are still imported
+# as `databricks.sdk.core.X` by callers.
+# ruff: noqa: F403, F405
+
 import re
 from typing import BinaryIO
 from urllib.parse import urlencode
 
 from ._base_client import _BaseClient
 from .config import *
-
-# To preserve backwards compatibility (as these definitions were previously in this module)
 from .credentials_provider import *
 from .errors import DatabricksError, _ErrorCustomizer
 from .oauth import retrieve_token

databricks/sdk/core.py

@@ -1,3 +1,6 @@
+# Re-exports the public errors API; star imports from .platform and .sdk are intentional.
+# ruff: noqa: F401, F403, F405, F811
+
 from .base import DatabricksError, ErrorDetail
 from .customizer import _ErrorCustomizer
 from .parser import _Parser

databricks/sdk/errors/__init__.py

@@ -1 +1,3 @@
+# ruff: noqa: F401
+
 from .round_trip_logger import RoundTrip

databricks/sdk/errors/overrides.py

@@ -124,7 +124,7 @@ def inner() -> Dict[str, str]:
         logging.debug(f"Failed to initialize globals 'sqlContext' and 'table', continuing. Cause: {e}")
 
     try:
-        from pyspark.sql.functions import udf  # type: ignore
+        from pyspark.sql.functions import udf  # type: ignore  # noqa: F401
     except ImportError as e:
         logging.debug(f"Failed to initialise udf global: {e}")