What this PR does / why we need it:

_convert_value_type_str_to_value_type maps value type names to ValueType, falling back to STRING for unknown names. Three names fall back by mistake:

• the INT32_LIST key is written as "INT32_LIST " (trailing space), so "INT32_LIST" never matches,
• PDF_BYTES and IMAGE_BYTES are not in the map.

from feast.type_map import _convert_value_type_str_to_value_type
_convert_value_type_str_to_value_type("INT32_LIST")   # ValueType.STRING
_convert_value_type_str_to_value_type("PDF_BYTES")    # ValueType.STRING
_convert_value_type_str_to_value_type("IMAGE_BYTES")  # ValueType.STRING

The registry REST API uses this function to turn a field's valueType into a Feast type (api/registry/rest/feature_views.py, features.py, data_sources.py), so Array(Int32), PdfBytes and ImageBytes features are reported as String, including in the generated Field(name=..., dtype=...) code. The Snowflake UDF helper create_entity_dict uses it too.

This removes the trailing space and adds the two missing names.

Which issue(s) this PR fixes:

No issue, found while checking type name conversions for each ValueType.

Checks

• I've made sure the tests are passing.
• My commits are signed off (git commit -s)
• My PR title follows conventional commits format

Testing Strategy

• Unit tests
• Integration tests

Added test_convert_value_type_str_round_trips_every_name in sdk/python/tests/unit/test_type_map.py, which checks every ValueType name; the INT32_LIST, PDF_BYTES and IMAGE_BYTES cases fail on master. The other tests of this function pass. ruff check and ruff format --check are clean on the changed files.