What this PR does / why we need it:

A None inside a UUID_LIST, TIME_UUID_LIST or DECIMAL_LIST is written to the proto as "" (_LIST_NONE_DEFAULTS in type_map.py, because a repeated string field cannot hold None). feast_value_type_to_python_type then calls UUID("") or Decimal("") on it. So the write works and the read fails:

fv = FeatureView(..., schema=[Field(name="u", dtype=Array(Uuid)), Field(name="dec", dtype=Array(Decimal))], ...)
store.apply([entity, fv])
df = pd.DataFrame({"id": [1], "event_timestamp": [...], "u": [[str(u), None]], "dec": [["1.5", None]]})
store.write_to_online_store("fv", df)
store.get_online_features(features=["fv:u", "fv:dec"], entity_rows=[{"id": 1}]).to_dict()
# raises ValueError: badly formed hexadecimal UUID string
# or decimal.InvalidOperation: [<class 'decimal.ConversionSyntax'>],
# depending on which column is converted first

With this change the same call returns {'u': [[UUID('...'), None]], 'dec': [[Decimal('1.5'), None]]}. The reader now maps "" to None for the three list types, the way the unix_timestamp_list_val branch maps its NULL_TIMESTAMP_INT_VALUE sentinel to None. The scalar and set branches are not touched.

Cause: the list readers and the "" default.

Which issue(s) this PR fixes:

No issue, found while checking list conversions in type_map.py.

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_uuid_list_with_none_roundtrip, test_time_uuid_list_with_none_roundtrip and test_decimal_list_with_none_roundtrip to sdk/python/tests/unit/test_type_map.py. All three fail on master with the errors above and pass with the fix: pytest sdk/python/tests/unit/test_type_map.py -k with_none_roundtrip gives 3 passed. The rest of the file passes apart from TestSparkNativeTypeValidation (needs pyspark). ruff check and ruff format --check are clean on both files.