[Artmann]

Adds GET /api/v1/datapoints to make it easier to see what is stored for a project when debugging.

What it does

• Lists datapoints read straight from S3, so it also works while the index is still loading.
• Pagination with limit (1 to 200, default 50) and an opaque cursor; the response carries nextCursor, null on the last page.
• Filters: sourceId (direct lookup, no scan), audience (project means none), label (case-insensitive substring), occurredFrom / occurredTo.
• includeContent=true adds content; it is off by default since it costs an extra read per datapoint.
• Each item has the stored fields plus contentTokens, contentSha256, embeddingModel and embeddingDim, without the vector.

Trade-offs

• Pages are in id order, so S3 can list them incrementally with StartAfter, rather than sorted by time.
• One request reads at most 1000 datapoints. With narrow filters a page can be short or empty but still have a nextCursor.

Docs

README: documents the endpoint, adds the seed script to the quick start, fixes which endpoints return 503 while loading, and refreshes the status and "Not yet" sections.

Testing

• cargo test --test datapoints -- --ignored against LocalStack: paging, each filter, listing while loading, validation.
• Existing delete and recall suites, unit tests, clippy and fmt pass.

🤖 Generated with Claude Code