[ump45nose]

Part of #4078 (the "Avro timestamp-nanos raises Unknown logical type" item).

Rationale for this change

AvroSchemaConversion.iceberg_to_avro already writes timestamp_ns / timestamptz_ns as {"type": "long", "logicalType": "timestamp-nanos", "adjust-to-utc": ...}, but avro_to_iceberg has no branch for timestamp-nanos. So an Avro file PyIceberg writes with a nanosecond timestamp field can't be opened by AvroFile, because it converts the schema from the file header first:

ValueError: Unknown logical/physical type combination: {'type': 'long', 'logicalType': 'timestamp-nanos', 'adjust-to-utc': False}

This change maps timestamp-nanos to TimestampNanoType, or to TimestamptzNanoType when adjust-to-utc is true, the same way timestamp-micros is handled. The readers and writers for these types already exist in pyiceberg/avro, so only the schema conversion was missing. I also fixed the parameter type hints on the two visit_timestamp*_ns methods, which said TimestampType / TimestamptzType.

Are these changes tested?

Yes:

• tests/utils/test_schema_conversion.py: timestamp-nanos with adjust-to-utc missing, false, and true, plus an Iceberg -> Avro -> Iceberg schema round trip.
• tests/avro/test_file.py: writes a file with timestamp_ns and optional timestamptz_ns fields using AvroOutputFile and reads it back with AvroFile.

All 5 new tests fail on main and pass with this change. pytest tests/avro tests/utils/test_schema_conversion.py passes (193 tests), and ruff check/format are clean.

Are there any user-facing changes?

Avro files with timestamp-nanos fields can now be read. No API changes.

---

AI disclosure: I used an AI coding assistant to help write this change and its tests. I reviewed the code and ran the tests myself.