docs(explanation): add caching chapter covering all layers

[jensens]

New Diataxis Explanation chapter at `docs/sources/explanation/caching.md`.

Why

We had caching fragments scattered across `performance.md` but no consolidated page that answers "what is cached where, for how long, and how does it get invalidated" — which is exactly what a reader needs when sort/pagination looks unstable or `getObject()` appears to return stale data.

What it covers

• All six cache layers on the query → getObject path (query result cache, prepared statements, request connection pool, zodb-pgjsonb `LoadCache`, ZODB Connection object cache, PG `shared_buffers`) with scope/lifetime/invalidation.
• The explicit invariant: brains and `getObject()` results are deliberately not memoized. Why that matters for request-boundary hygiene.
• Prefetch mechanism (`_maybe_prefetch_objects` + `storage.load_multiple()`) — what it warms, what it doesn't.
• Invalidation matrix: which write event clears which cache.
• Every configuration knob (env vars for plone.pgcatalog, `zope.conf` for ZODB, `` for the storage, `postgresql.conf` for PG).
• Debugging guide for the three common "my cache isn't behaving" symptoms.
• Cross-links to `performance.md` for benchmarks (no duplication) and `architecture.md` for the write path that drives catalog-side invalidation.

Test plan

• `sphinx-build -W` clean (the two unrelated TOC warnings are pre-existing and about other how-to pages)
• Vale clean (added `memoize`, `unpickle`, `truthy`, `unsortable` to project vocab)
• Mermaid diagram renders
• Chapter added to `explanation/index.md` TOC between `tika-extraction` and `performance`

🤖 Generated with Claude Code