Avro: Cache Avro schema conversion in AvroFileHeader.get_schema (#3662)

[hedger9487]

Description

Fixes #3662.

Every manifest file under a partition spec embeds an identical Avro schema string (avro.schema). Previously, AvroFileHeader.get_schema() in pyiceberg/avro/file.py performed a full json.loads() and recursive AvroSchemaConversion().avro_to_iceberg() conversion on every manifest read during scan planning.

This PR adds an LRU cache for Avro schema string parsing via _parse_avro_schema, avoiding redundant conversions across repeated manifest reads during scan planning (e.g. accelerating scan().plan_files() by ~1.8x on tables with many manifests).

Testing

• Added unit test test_get_schema_is_cached in tests/avro/test_file.py.
• All 49 Avro and manifest tests pass locally.

[Copilot]

Pull request overview

This PR optimizes Avro manifest/schema handling by caching the Avro-schema-string → Iceberg Schema conversion, avoiding repeated json.loads() + recursive conversion work across many manifest reads during scan planning.

Changes:

• Added an @lru_cache(maxsize=128)-backed _parse_avro_schema() helper to cache schema parsing/conversion by schema string.
• Updated AvroFileHeader.get_schema() to use the cached helper.
• Added a unit test asserting schema conversion results are reused (object identity) across headers with the same embedded schema string.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
pyiceberg/avro/file.py	Introduces an LRU-cached schema parsing/conversion helper and routes get_schema() through it.
tests/avro/test_file.py	Adds a regression test to ensure schema conversion is cached and reused across calls.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.