fix(mcp): keep search index type in vector hydration

[rudi193-cmd]

Fixes #982.

Summary

• Key vector-search hydration by (type, id) instead of bare id, so entity/relation/observation rows with the same search_index id no longer overwrite each other.
• Keep the existing broad WHERE id IN (...) query shape and apply type discrimination in Python, matching the existing filtered-search guard.
• Update vector-search test mocks and add a regression test covering entity:4:0 and relation:4:0 both hydrating successfully.

Test plan

• uv run pytest tests/repository/test_vector_threshold.py tests/repository/test_vector_pagination.py
• uv run ruff format src/basic_memory/repository/search_repository_base.py tests/repository/test_vector_threshold.py tests/repository/test_vector_pagination.py
• uv run ruff check src/basic_memory/repository/search_repository_base.py tests/repository/test_vector_threshold.py tests/repository/test_vector_pagination.py

[rudi193-cmd]

Closing in favor of #984, which contains the same patch with the required DCO sign-off trailer.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 16144196fc

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[chatgpt-codex-connector]

Preserve row type during hybrid fusion

Now that vector hydration can append multiple SearchIndexRows with the same numeric id but different type, default hybrid searches still drop one of them because _search_hybrid stores FTS/vector rows in rows_by_id, fts_scores, and vec_scores keyed only by row.id (see src/basic_memory/repository/search_repository_base.py around lines 2207-2239). In semantic-enabled clients where hybrid is the default, a vector match for entity:4 and relation:4 will collapse to whichever row is assigned last, so the collision this patch fixes for vector-only remains user-visible in hybrid mode; carry the (type, id) key through the fusion maps as well.

Useful? React with 👍 / 👎.