fix(semantic-index): make insertOne idempotent to survive mid-batch crash

With API batched embeddings, insertChunks() and stateStore.updateProgress()
write to two separate SQLite databases, so they cannot be atomic. If the
worker is terminated between the two writes, the next resume attempts to
re-insert the same chunk_ids and hits the UNIQUE constraint, leaving the
session stuck in 'failed' state.

Use INSERT OR IGNORE on chunk_vector_index and skip the vec-table insert
when changes === 0. Duplicate inserts during resume are now silently
discarded instead of throwing.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

packages/node-runtime/src/semantic-index/store.ts

@@ -126,7 +126,8 @@ export class EmbeddingIndexStore {
 
     const meta = this.db
       .prepare(
-        `INSERT INTO chunk_vector_index (${CHUNK_COLUMNS})
+        // ponytail: OR IGNORE makes resume after mid-batch crash idempotent; vec insert skipped below if chunk exists
+        `INSERT OR IGNORE INTO chunk_vector_index (${CHUNK_COLUMNS})
          VALUES (@chunkId, @dbPathHash, @strategyId, @modelId, @dim, @parentId, @startMessageId, @endMessageId,
                  @startTs, @endTs, @messageCount, @rawContentHash, @embeddingInputHash, @chunkerVersion,
                  @chunkerConfigHash, @indexedAt, @status)`
@@ -151,6 +152,7 @@ export class EmbeddingIndexStore {
         status: record.status,
       })
 
+    if (meta.changes === 0) return
     this.db
       .prepare(
         `INSERT INTO ${vecTableName(record.dim)} (vector_id, db_path_hash, model_id, embedding)