SourceTags should be dict[str, str], not dict[str, Any]

Summary

SourceTags is typed dict[str, Any], and it should be dict[str, str].

SourceTags = NewType("SourceTags", Annotated[dict[str, Any], Field(description=...)])
# packages/overture-schema-theme-base/src/overture/schema/base/_common.py:38

The pre-Pydantic hand-written YAML had:

sourceTags:
  description: Any attributes/tags from the original source data that should be passed through.
  type: object

type: object with no additionalProperties is JSON Schema's unconstrained object, which is what dict[str, Any] renders faithfully.

Why str is right

• Source tags are OSM tags, which are string-valued. Every source_tags example in the repo
  is string-valued: natural: dune, surface: sand, leisure: pitch, water: ocean, waterway: river.
• There is no coercion anywhere to support a wider type — no validator on SourceTags or
  source_tags, in either the Pydantic model or the old YAML.
• It is not an output change. PySpark already emits
  MapType(StringType(), StringType(), True) for this field, so narrowing the model makes it match what is already being produced rather than altering it.

There is also reason to think no non-string value is being written today: the PySpark schema already declares MapType(StringType(), StringType(), True), and Spark enforces the declared schema on write, so a non-string value would have failed rather than passed through. Narrowing the model records what the data already has to satisfy.

Start at packages/overture-schema-theme-base/src/overture/schema/base/_common.py:38 and inspect the SourceTags definition and related Pydantic model usage. Compare it with the repository's source_tags examples and the existing PySpark MapType(StringType(), StringType(), True) schema. Done means the model type records string-valued tags without changing the emitted schema.