What this PR does / why we need it:

ValueType.PDF_BYTES is missing from PYTHON_SCALAR_VALUE_TYPE_TO_PROTO_VALUE in type_map.py, while IMAGE_BYTES is there. So python_values_to_proto_values raises for any PdfBytes field, and a feature view with one can't be materialized or written to the online store:

fv = FeatureView(
    name="docs",
    entities=[doc],
    schema=[Field(name="doc_id", dtype=String), Field(name="pdf", dtype=PdfBytes)],
    source=FileSource(path=path, timestamp_field="event_timestamp"),
)
store.apply([doc, fv])
store.write_to_online_store("docs", df)   # same for store.materialize_incremental(...)
# Exception: Unsupported data type: ValueType.PDF_BYTES

The same feature view with ImageBytes instead works. This adds PDF_BYTES to the map with the same bytes_val conversion as BYTES and IMAGE_BYTES. Reading back already worked, since feast_value_type_to_python_type only looks at the proto field.

Which issue(s) this PR fixes:

No issue, found while checking value 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_python_values_to_proto_values_binary in sdk/python/tests/unit/test_type_map.py for BYTES, PDF_BYTES and IMAGE_BYTES; the PDF_BYTES case fails on master. I also ran the example above against a local SQLite online store: write and materialize both work and get_online_features returns the bytes. pytest tests/unit/test_type_map.py passes apart from the TestSparkNativeTypeValidation cases, which fail the same way on master here because pyspark isn't installed. ruff check and ruff format --check are clean on the changed files.