Show any-article nearest neighbours and widen the format/kind vocabularies

The article dashboard page gains a "Nearest articles (any)" table: the
ten closest stored embeddings by cosine, regardless of rating or run,
via a brute-force scan of article_embeddings. The deep-assessment format
facet grows from 5 to 14 values (code_repository, documentation_reference,
tool_or_product_page, discussion_thread, paper_or_report,
interview_or_transcript, video_or_podcast, fiction_or_humor, other) and
the triage kind from 10 to 16 (repo, docs, discussion, paper, media,
fiction), so a GitHub repository is no longer forced into analysis_essay.
Both prompt versions bump to 2.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QVPagF6jfDv78CC5Jv2wp4

tests/fixtures/deepseek_deep_batch_messy.json

@@ -26,7 +26,7 @@
       "rationale": "mailing-list argument about tape drives, oddly gripping",
       "paywalled_guess": "false",
       "facets": {
-        "format": "discussion_thread",
+        "format": "interactive_experience",
         "depth": "standard",
         "evidence": "anecdote",
         "commerciality": "none",

tests/m3_curation.rs

@@ -76,7 +76,7 @@ fn deep_messy_fixture_is_salvaged_not_rejected() {
     assert!(raw.contains("\"id\": \""), "needs a string id");
     assert!(raw.contains("\"quality\": \""), "needs a string score");
     assert!(
-        raw.contains("discussion_thread"),
+        raw.contains("interactive_experience"),
         "needs an unknown facet token"
     );