Add semantic pre-filter for GraphRAG edge scoring (#702)

Embed edge descriptions and compute cosine similarity against grounding
concepts to reduce the number of edges sent to expensive LLM scoring.
Controlled by edge_score_limit parameter (default 30), skipped when edge
count is already below the limit.

Also plumbs edge_score_limit and edge_limit parameters end-to-end:
- CLI args (--edge-score-limit, --edge-limit) in both invoke and service
- Socket client: fix parameter mapping to use hyphenated wire-format keys
- Flow API, message translator, gateway all pass through correctly
- Explainable code path (_question_explainable_api) now forwards all params
- Default edge_score_limit changed from 50 to 30 based on typical subgraph
  sizes

trustgraph-base/trustgraph/messaging/translators/retrieval.py

@@ -84,6 +84,7 @@ class GraphRagRequestTranslator(MessageTranslator):
             triple_limit=int(data.get("triple-limit", 30)),
             max_subgraph_size=int(data.get("max-subgraph-size", 1000)),
             max_path_length=int(data.get("max-path-length", 2)),
+            edge_score_limit=int(data.get("edge-score-limit", 30)),
             edge_limit=int(data.get("edge-limit", 25)),
             streaming=data.get("streaming", False)
         )
@@ -97,6 +98,7 @@ class GraphRagRequestTranslator(MessageTranslator):
             "triple-limit": obj.triple_limit,
             "max-subgraph-size": obj.max_subgraph_size,
             "max-path-length": obj.max_path_length,
+            "edge-score-limit": obj.edge_score_limit,
             "edge-limit": obj.edge_limit,
             "streaming": getattr(obj, "streaming", False)
         }