eval: GPU rerank baseline + CLI fix

GPU eval (hybrid+rerank, RERANK_URL=http://10.10.1.65:8082):
  MRR=0.672  Recall@5=0.638  nDCG@5=0.621  (35 queries, 1 transient
  500, otherwise clean)

Quality identical to the CPU rerank run as expected — only latency
changed (single rerank call dropped from ~23s to ~0.7-1.5s on the
Tesla P4). Per-query report at eval/results/with_rerank_gpu.md.

CLI parser fix: `--retrievers dense+rerank,hybrid+rerank` now
correctly wires the dense+rerank variant. Previously only literal
"rerank" (without prefix) matched the dense+rerank branch, so
combined-retriever runs silently dropped dense+rerank.

(Note: the eval's RerankedRetriever does 50 individual Chroma `get`
calls per query to fetch chunk text by (source, source_key); this
adds ~15s per query of pure SQLite lookup overhead. Not a production
concern — docs_mcp/server.py's _rerank_pool reranks docs already in
the dense pool, no extra Chroma round-trips. Worth tightening the
eval-side impl on a later pass.)

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

eval/run_eval.py

@@ -102,7 +102,7 @@ def main() -> int:
         DenseRetriever, BM25Retriever, HybridRetriever, RerankedRetriever
     )
 
-    wanted = [x.strip() for x in args.retrievers.split(",") if x.strip()]
+    wanted = {x.strip() for x in args.retrievers.split(",") if x.strip()}
     dense = DenseRetriever()
     bm25 = BM25Retriever()
 
@@ -113,7 +113,8 @@ def main() -> int:
         retrievers.append(("bm25", bm25))
     if "hybrid" in wanted:
         retrievers.append(("hybrid-rrf", HybridRetriever(dense=dense, bm25=bm25, pool=args.pool)))
-    if "rerank" in wanted:
+    # Accept either "rerank" or "dense+rerank" for the dense-base reranker.
+    if "rerank" in wanted or "dense+rerank" in wanted:
         retrievers.append(("dense+rerank",
                            RerankedRetriever(base=dense, pool=args.pool)))
     if "hybrid+rerank" in wanted: