test: Added reference parquet files and tests for ObjectCodec implementations

[malhotrashivam]

Closes #5767

[rcaudy]

Looks good. Please add coverage for more null cases (easy just to have a null for every type), negative values in BigDecimalCodec, compact values in LocalTimeCodec, and ExternalizableCodec.

[malhotrashivam]

Looks good. Please add coverage for more null cases (easy just to have a null for every type), negative values in BigDecimalCodec, compact values in LocalTimeCodec, and ExternalizableCodec.

• Added null values in columns, but they still don't cover the null cases in codec because these values are encoded as parquet nulls and not as null objects. Therefore, we don't hit the null case in codec.
• Added negative values in BigDecimalCodec and compact values in LocalTimeCodec
• ExternalizableCodec::decode was already covered in TestCodecColumns.

[rcaudy]

Looks good. Please add coverage for more null cases (easy just to have a null for every type), negative values in BigDecimalCodec, compact values in LocalTimeCodec, and ExternalizableCodec.

• Added null values in columns, but they still don't cover the null cases in codec because these values are encoded as parquet nulls and not as null objects. Therefore, we don't hit the null case in codec.
• Added negative values in BigDecimalCodec and compact values in LocalTimeCodec
• ExternalizableCodec::decode was already covered in TestCodecColumns.

• Our legacy Parquet behavior didn't properly encode nulls, right?
• If ExternalizableCodec was covered, why didn't it show up on my coverage report? My coverage report was lying.

[malhotrashivam]

I tried regenerating data before my change #4191 but that didn't hit the null case in codec too because that issue was only in the primitive types where, for example, we stored null value as NULL_CHAR or NULL_INT instead of parquet nulls.