Fix: use 'document' prompt for index embeddings, not 'query'

Qwen3-Embedding-0.6B has separate prompt templates for queries vs
documents. Documents should use 'document' prompt at index time,
'query' prompt at search time. Was using 'query' for both.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

build_embeddings.py

@@ -88,7 +88,7 @@ def embed_datasets(model, dataset_source):
         summaries,
         batch_size=BATCH_SIZE,
         show_progress_bar=True,
-        prompt_name="query",
+        prompt_name="document",
     )
     logger.info("Embeddings computed")
 
@@ -125,7 +125,7 @@ def embed_models(model, model_source):
         summaries,
         batch_size=BATCH_SIZE,
         show_progress_bar=True,
-        prompt_name="query",
+        prompt_name="document",
     )
     logger.info("Embeddings computed")