feat: Croissant ML dataset discovery via federation-native PAP agents

[toadkicker]

Summary

• Federation-native dataset search: Two new PAP agents (schema:DatasetAction) for HuggingFace Dataset Search and OpenML Dataset Search — pure TOML, zero new Rust trait code, auto-discovered by FederatedRegistry.query_local_satisfiable()
• Parallel JoinSet handshake fan-out: All registered zero-disclosure dataset agents run the full 6-phase PAP handshake concurrently; results merged by preference-biased relevance score
• Long-horizon memex integration: FTS5 prior-art check with 7-day TTL emits RestoredFromMemex immediately while fresh handshakes run in parallel; per-agent and aggregate episodes recorded; PreferenceEngine signals fed back for future ranking
• DatasetSearchTemplate block renderer: Reads schema:ItemList JSON-LD, shows memex provenance banner, source-agent badges, open-license tint, and receipt provenance footer
• Intent expansion: 47 tests (up from 32), new schema:DatasetAction rule, free-dictionary coverage, strip-behaviour and ordering-conflict regression tests
• BlockUpdate migration: Adapted all 6 emit sites to the BlockUpdate wire type from refactor: BlockUpdate typed events + BlockContext provider pattern #289 (reviewed the refactor before applying — it's correct: compiler-enforced separation of backend vs. frontend-only state)

Architecture notes

Adding a new dataset source = add one TOML file. No Rust changes required. Federation self-configuration: publish_agent(did, "pap://shared-registry") makes any Papillon instance auto-discover it via FederatedRegistry.merge_remote().

Test plan

• cargo test --workspace passes (271 shared-crate tests, all intent tests including 15 new ones)
• cargo clippy --workspace --all-targets -- -D warnings clean
• cargo fmt --all -- --check clean
• cargo check --target wasm32-unknown-unknown clean in apps/papillon/frontend
• CI smoke-test (trunk WASM build + Playwright) passes
• CI chrysalis E2E passes

🤖 Generated with Claude Code

[github-actions]

Benchmark Regression Report

PAP Protocol Benchmark Regression Check
========================================
Baseline: .bench-baseline/baseline.json
Threshold: 30%

  ed25519_keypair_generation                19.7 µs  (baseline: 19.7 µs, +0.0%)  [ok]
  did_key_derivation                         1.5 µs  (baseline: 1.5 µs, +0.0%)  [ok]
  mandate_create_sign                       24.0 µs  (baseline: 24.0 µs, -0.2%)  [ok]
  mandate_chain_verify_depth3              125.1 µs  (baseline: 126.0 µs, -0.7%)  [ok]
  sd_jwt_issue_5claims                      28.0 µs  (baseline: 27.9 µs, +0.3%)  [ok]
  sd_jwt_verify_disclose_3of5               44.1 µs  (baseline: 44.5 µs, -1.0%)  [ok]
  session_open_full_lifecycle              111.9 µs  (baseline: 108.6 µs, +3.0%)  [ok]
  receipt_create_cosign                     47.8 µs  (baseline: 49.0 µs, -2.5%)  [ok]
  federation_announce_local                 55.9 µs  (baseline: 55.9 µs, +0.0%)  [ok]

All benchmarks within 30% of baseline.

_{Threshold: 10% regression vs baseline from main}