What this PR does / why we need it:

feast materialize crashes with ValueError: The truth value of an empty array is ambiguous when a scalar feature column contains an empty numpy array (e.g. np.array([])). This is a real-world scenario when a DataFrame row has a missing value represented as an empty array rather than None or np.nan.

Root cause: In _convert_scalar_values_to_proto (sdk/python/feast/type_map.py), the null check uses not pd.isnull(value) for every value in the loop. pd.isnull() is vectorised — when value is a numpy array, it returns a boolean array instead of a scalar. Applying Python's not operator to that array raises ValueError. The same issue exists in:

• the BOOL scalar path (not pd.isnull(value) in a list comprehension)
• the UNIX_TIMESTAMP early-return path (_python_datetime_to_int_timestamp(values) called with the raw values list, including any array-like values)
• the sample type-validation check (sample == 0)

Fix: Before calling pd.isnull(), guard both scalar conversion loops (generic and BOOL) and the sample type-validation with an explicit isinstance(value, np.ndarray) check. Any array-like value in a scalar feature column is unmappable to a protobuf scalar field anyway, so it is safely treated as null → ProtoValue().

Which issue(s) this PR fixes:

Fixes #6255

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
• Manual tests
• Testing is not required for this change

Misc