fix(dispatch): don't count gate-failing implementers toward confidence

_dispatch_confidence was computed from the raw implementers set before
the per-implementer language gate ran, so a sourceless/external-reference
implementer, or a cross-language name collision, that never produces an
edge could still dilute the confidence score for the implementer(s) that
do. A single real implementer alongside one gate-failing entry scored
0.75 (the two-implementer tier) instead of 0.85.

Gate first, then count only the implementers that pass it, so the score
reflects how many implementers actually share the dispatch edge.

Addresses review feedback on PR #4396 (Graphify review bot, finding at
interface_dispatch.py:154).

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
(cherry picked from commit b01a6fcf10)

tests/test_csharp_interface_dispatch.py

@@ -266,6 +266,37 @@ def test_a_non_csharp_implementer_is_not_dispatched_to(tmp_path):
     assert not dispatch
 
 
+def test_a_gate_failing_implementer_does_not_dilute_confidence(tmp_path):
+    # `implementers[iface]` holds every `implements` source regardless of
+    # language or whether the node even resolved (a dangling/external
+    # reference stub). Only `A` passes the gate and gets an edge; the stub
+    # must not count toward "how many implementers" for confidence purposes,
+    # or a single real implementer would score as if it had a sibling.
+    from graphify.csharp_dispatch import resolve_csharp_interface_dispatch
+
+    nodes = [
+        {"id": "iface", "label": "IR", "source_file": "a.cs", "_callable_class": True},
+        {"id": "a", "label": "A", "source_file": "a.cs", "_callable_class": True},
+        {"id": "stub", "label": "Stub", "_callable_class": True},  # sourceless: fails declared_in
+        {"id": "iface_m", "label": ".M()", "source_file": "a.cs", "_callable": True},
+        {"id": "a_m", "label": ".M()", "source_file": "a.cs", "_callable": True},
+        {"id": "stub_m", "label": ".M()", "_callable": True},
+    ]
+    edges = [
+        {"source": "a", "target": "iface", "relation": "implements"},
+        {"source": "stub", "target": "iface", "relation": "implements"},
+        {"source": "iface", "target": "iface_m", "relation": "method"},
+        {"source": "a", "target": "a_m", "relation": "method"},
+        {"source": "stub", "target": "stub_m", "relation": "method"},
+    ]
+    resolve_csharp_interface_dispatch([], nodes, edges)
+    dispatch_edges = [e for e in edges if e["relation"] == "dispatches_to"]
+    assert len(dispatch_edges) == 1
+    assert dispatch_edges[0]["source"] == "iface_m"
+    assert dispatch_edges[0]["target"] == "a_m"
+    assert dispatch_edges[0]["confidence_score"] == 0.85
+
+
 def test_mixed_corpus_links_only_the_csharp_pair(tmp_path):
     dispatch, r = _extract(tmp_path, {
         "IReport.cs": "public interface IReport { void Build(); }\n",