feat: motion content enrichment pipeline hardening

- ai_provider_wrapper: retry/fallback with exponential backoff, None sentinel for failed items
- text_pipeline: use wrapper, return 5-tuple (stored, skipped_existing, skipped_no_text, errors, failed_ids)
- similarity/compute: filter trivial 1.0 matches on identical short titles (<12 chars)
- rerun_embeddings: --retry-missing mode, calls ensure_text_embeddings_for_ids on failed ids
- sync_motion_content: per-ext_id retries, HTTPAdapter pool, --max-body-workers CLI flag, audit on failure
- qa_similarity script: samples motions, writes JSON ledger to thoughts/ledgers/
- All tests green: 61 passed, 2 skipped

pipeline/text_pipeline.py

@@ -177,11 +177,7 @@ def ensure_text_embeddings(
         )
 
     skipped_existing = int(existing)
-    # Historically some callers expected a 4-tuple; return the primary
-    # metrics (stored, skipped_existing, skipped_no_text, errors).
-    # The list of failed_ids is intentionally not returned here to remain
-    # backward-compatible with older callers.
-    return stored, skipped_existing, skipped_no_text, errors
+    return stored, skipped_existing, skipped_no_text, errors, failed_ids
 
 
 def ensure_text_embeddings_for_ids(