Next steps after initial implementation

[tomcrane]

All 10 PRs in the original suggested sequence are now complete. This issue tracks potential next steps for further refinement and extension.

Possible next steps

• IIIF Search API v2 — implement the v2 search and autocomplete endpoints alongside the existing v1 endpoints
• hOCR format provider — add an ITextFormatProvider implementation for hOCR alongside the existing AltoTextFormatProvider
• S3 deployment wiring — wire S3TextStore into the Builder and Search API configurations; add infrastructure/deployment configuration for AWS
• Production observability — structured logging, metrics, health checks; guidance on what to monitor in production (see also Search API cache: memory sizing strategy for large Text objects #9 for cache sizing)
• IIIF Presentation v2 Manifest support — the Builder API currently rejects non-v3 Manifests; v2 support may be needed for some collections
• Comparative validation — run the Search API against known-good Wellcome search results to validate word positions and bounding boxes
• Performance testing — load test Search API with large Text objects; validate cache behaviour and LOH pressure under realistic traffic (see instructions/cache-usage.md) (moved to Load / Perf testing of search #60)
• Playwright / browser E2E — now that the API E2E suite exists, browser-level tests for any future UI components

This is a backlog for discussion and prioritisation.