[Arshdeep54]

Closes #99

Points can now carry a sparse vector alongside a dense one. Sparse vectors go into a new in-memory inverted index and are searched by dot product.

• Search now takes a single query that is either dense or sparse, and returns hits as { id, score }. For dense queries the score is a distance (lower is closer); for sparse it's a dot product (higher is closer).
• Search limit is now capped at 10,000.
• HTTP, gRPC, the Python client, the TUI and the docs are updated to the new shapes.

Also, the stored point format changed, so existing data directories and snapshots won't load with this version. Start from a fresh data directory and re-insert.

[mintlify]

Preview approval required

This pull request comes from a fork, so Mintlify won't build a preview until someone with write access approves commit 0817d5a.

Approve preview