Reject extra keys schema validation

[Scr4tch587]

What changed

Added an extra-key check to validate() in core/runtime/validator.py. Before inserting builder output, the
validator now raises ValidationError if a row contains any key not declared in the dataset's schema.

Added schema={"val": SchemaType.INT} to 4 worker unit tests whose mock output was silently passing the old
empty-schema check.

Why

Extra keys were silently dropped, which made schema drift hard to catch: builders could diverge from their
declared schema without any visible error.

Testing

1. Unit, updated test_validator.py: replaced the old test_extra_keys_allowed (which asserted the wrong behaviour) with test_extra_keys_raises, confirming a ValidationError is raised with the expected message.
2. Integration, added test_schema_unexpected_key to test_errors.py: builder returns a valid schema key plus
   an extra key, confirms the build returns 500 and no rows are inserted.
3. All 13 unit tests and 38 integration tests pass.

[github-actions]

• kai/extra-keys-schema-validation - ⚠️ No PR associated with branch
  • Reject extra keys schema validation 👈
    • fix: validate builder row output type #347