feat(aws-dataprocessing): add Iceberg format detection, Data Quality, table versions, and partition index tools

[dipayanthedata]

Summary

Related to RFC #4341.

aws-dataprocessing-mcp-server already covers the vast majority of the AWS Glue Data Catalog surface area, but four specific gaps were verified to be completely absent from the codebase, tests, README, and git history. This PR closes all four, following the server's existing "mega-tool with an operation parameter" convention rather than introducing a parallel design:

1. Iceberg/table-format detection — new detect-table-format and get-iceberg-table-info operations on manage_aws_glue_tables. Classifies a table as Hive, Iceberg, Delta, or Hudi from its Glue table parameters and storage descriptor (best-effort, falls back to UNKNOWN rather than guessing), and surfaces Iceberg metadata location / format version when applicable.
2. Glue Data Quality — a new manage_aws_glue_data_quality tool covering ruleset CRUD, starting/inspecting ruleset evaluation runs, and retrieving evaluation results (single, list, and batch-get). This also wires up the DataQualityResult Pydantic model in models/data_catalog_models.py, which previously existed but was never used by any handler — the README already advertised "data quality validation" as a feature before this PR, without any implementation behind it.
3. Table version history — new get-table-versions and batch-delete-table-version operations on manage_aws_glue_tables.
4. Partition indexes — new create-partition-index, get-partition-indexes, and delete-partition-index operations on manage_aws_glue_partitions.

Notable fix found during testing

While writing tests for get-partition-indexes, found that GetPartitionIndexes' Keys field in the real Glue API response is a list of {Name, Type} dicts, not plain strings. Fixed the manager code to extract key names correctly (awslabs/aws_dataprocessing_mcp_server/core/glue_data_catalog/data_catalog_handler.py).

Testing

• Added unit tests for every new operation: happy path, empty results, not-found, and permission-boundary (--allow-write) cases, using this server's existing unittest.mock.MagicMock/patch convention (not moto — kept consistent with every other test in this server rather than introducing a second testing style).
• Full suite: 1218 passed.
• ruff check . and ruff format --check .: clean, no changes needed beyond auto-formatting during development.

Compatibility

• No new dependencies added — everything uses boto3 Glue API calls already reachable from the existing glue_client.
• No changes to any existing operation's behavior — all new operations are additive to already-shipped tools (manage_aws_glue_tables, manage_aws_glue_partitions) or a new standalone tool (manage_aws_glue_data_quality).
• Write operations (batch-delete-table-version, create-partition-index, delete-partition-index, and the Data Quality ruleset/evaluation-run mutations) are gated behind --allow-write, consistent with every other mutating operation in this server.
• README updated: new tool/operation tables, IAM read-only and write policy JSON, and Key Features section.

Test plan for reviewers

• Review new operation docstrings for clarity/AWS permission accuracy
• Confirm IAM policy additions in README match the new API calls
• Optionally exercise detect-table-format / get-iceberg-table-info against a real Iceberg table to sanity-check the heuristic

---

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of the project license.

[github-actions]

This pull request is now marked as stale because it hasn't seen activity for a while. Add a comment or it will be closed soon. If you wish to exclude this issue from being marked as stale, add the "backlog" label.