fix(deploy): bump DATASET_CACHE_BUST -> ki138-v2 — force rebuild to pull post-KI-138 Chroma + extracted

The previous rebuild (24382536) was triggered by KI-140 commit 9c01288 with
DATASET_CACHE_BUST=ki137-v1. That rebuild's snapshot_download layer ran in
a race against two in-flight HF Dataset pushes:
1. Chroma binary re-upload (capturing KI-138's 27 in-place metadata updates)
2. Extracted JSONs re-upload (with KI-138 canonicalized names + slug
renames for the 2 irda->irdai cases)
Both pushes completed AFTER the rebuild started. To eliminate any ambiguity
about which snapshot the current container holds, bump cache_bust again so
the next rebuild is GUARANTEED to pull the now-final dataset state:
• rag/corpus/ 206 PDFs
• rag/extracted/ 201 canonicalized JSONs
• rag/vectors/ 7,317-chunk Chroma (KI-137 ingest + KI-138 metadata)

Dockerfile

@@ -70,7 +70,7 @@ COPY 40-data ./40-data
 # (7356 chunks from a prior ingest) instead of the freshly-uploaded
 # cleaned one (3799 chunks). Bump CACHE_BUST manually each time the
 # dataset is re-uploaded; the value just needs to change.
-ARG DATASET_CACHE_BUST=2026-05-15-ki137-v1
+ARG DATASET_CACHE_BUST=2026-05-15-ki138-v2
 RUN echo "Dataset cache bust: ${DATASET_CACHE_BUST}" && python -c "\
 from huggingface_hub import snapshot_download; \
 snapshot_download(\