feat(cli,agent): expose tiered search + taxonomy via CLI and wire into query/chat agent

[sebastianbraun25]

Problem

search_wiki (#233/#234, extended in #258/#259) is only reachable from inside the
openkb query/openkb chat agent. There is no way for a script, a CI job, or an external
AI assistant (e.g. GitHub Copilot, Claude Code) that isn't running the OpenKB agent itself
to browse the taxonomy (concepts/entities) or run the tiered summaries/sources search — the
only options today are re-implementing keyword search with grep, or paying for a full
openkb query LLM round-trip just to get a list of candidate pages.

Separately, the query/chat agent itself still navigates concepts/entities via index.md's
one-line summaries plus ad-hoc read_file calls, rather than the new, purpose-built
list_taxonomy_items/get_taxonomy_item (#259) — so it doesn't yet benefit from the
tiered search's precision/recall separation either.

Vorschlag

• CLI: openkb list-taxonomy [--kind concept|entity] [--json] and
  openkb search <query> [--scope briefs,summaries,sources] [--top-k N] [--json], so any
  external caller (script, CI, or an AI assistant's shell tool) gets the same taxonomy
  browsing and tiered search the agent uses, with zero LLM cost.
• Wire list_taxonomy/the retiered search_wiki into build_query_agent
  (openkb/agent/query.py, inherited by chat.py), replacing the old
  index.md-plus-ad-hoc-read_file path for concepts/entities and the single-tier
  search_wiki for summaries/sources, with updated search-strategy instructions.

Kontext

• Builds on feat(search): tiered BM25 search (briefs/summaries/sources) + taxonomy accessors #258/feat(search): tiered BM25 search (briefs/summaries/sources) + taxonomy accessors #259's TieredWikiSearch/list_taxonomy_items/get_taxonomy_item.
• No CLI/MCP/agent wiring existed for either capability before this — pure exposure of
  already-implemented, already-tested core logic.

---

_{This issue was drafted with the assistance of an AI assistant.}