feat(databases): add description field to fork lineage

[hotdata-automation]

Auto-generated from the updated HotData OpenAPI spec.
Source: https://github.com/hotdata-dev/www/pull/432

[claude]

The [Unreleased] entry does not describe this regeneration. This PR removes two public methods and adds four public methods, and neither group is recorded.

Removed from ConnectionsApi:

• purge_connection_cache (DELETE /v1/connections/{connection_id}/cache)
• purge_table_cache (DELETE /v1/connections/{connection_id}/tables/{schema}/{table}/cache)

Failure scenario: a user on 0.11.0 upgrades and calls connections_api.purge_connection_cache(cid). Python raises AttributeError: 'ConnectionsApi' object has no attribute 'purge_connection_cache'. The CHANGELOG offers no explanation, so the user cannot tell an SDK bug from an intentional API retirement.

The 0.11.0 entry records the equivalent endpoint retirements under a ### Removed heading with a **Breaking:** prefix. Follow that convention here.

Also add the new surface under ### Added:

• DatabasesApi.get_database_lineage, DatabasesApi.lookup_database_by_name, DatabasesApi.set_database_table_constant_per_key
• ConnectionsApi.set_managed_table_constant_per_key
• models DatabaseLineageResponse, ForkedFromInfo, LineageAncestorInfo, LineageForkInfo, ManagedTableConstantPerKeyResponse, UpdateManagedTableRequest
• LoadManagedTableRequest.mode now accepts delete, update, and upsert

[claude]

Review

Context disclosure: the full diff was omitted from the review prompt because of its size. gh pr diff 199 returned 460 KB, too large to read in one pass. Review is based on that diff file searched by path and hunk, plus direct reads of CHANGELOG.md, hotdata/models/database_lineage_response.py, hotdata/models/forked_from_info.py, hotdata/models/lineage_fork_info.py, hotdata/models/update_managed_table_request.py, hotdata/models/managed_table_constant_per_key_response.py, hotdata/api/databases_api.py, hotdata/api/connections_api.py, tests/integration/test_connections_read.py, tests/integration/test_database_fork.py, tests/integration/test_managed_tables_lifecycle.py, .github/workflows/integration-tests.yml, and .github/workflows/check-release.yml. Every new generated model was read in full.

The generated code itself is consistent. Both blocking issues concern surrounding files that the regeneration left behind.

Blocking Issues

1. CHANGELOG.md:12 — the entry omits two breaking removals and four new endpoints. See the inline comment. This PR deletes ConnectionsApi.purge_connection_cache and ConnectionsApi.purge_table_cache. The CHANGELOG records only feat(databases): add description field to fork lineage.

2. tests/integration/test_connections_read.py:32 — the test calls a method this PR deletes. Line 32 calls connections_api.purge_connection_cache(connection_id). hotdata/api/connections_api.py no longer defines that method. Failure scenario: the @pytest.mark.skip on line 15 hides this today, so the Integration Tests / integration job stays green. The skip reason says to re-enable once the seeded connection is restored. On that day the test raises AttributeError: ConnectionsApi object has no attribute purge_connection_cache instead of testing the connection. Fix: delete lines 31-32, and update the module docstring on line 4, which still lists cache purge as covered behavior.

Action Required

1. Rewrite the [Unreleased] CHANGELOG section. Add a Removed block marked **Breaking:** for the two purge methods, matching the 0.11.0 entry. Add an Added block for the new endpoints and models.
2. Remove the purge_connection_cache call and the stale docstring line from tests/integration/test_connections_read.py.

Note on tests/integration/test_managed_tables_lifecycle.py:17: the docstring mentions a purge_table_cache step that the scenario skips. No code depends on it, so it is safe to leave, but it now names a method the SDK does not expose.