fix(eval): parse_anomaly_segments must tolerate non-numeric junk entries

Full-scale eval crashed in the pure-LLM QA-judge path: pure_llm emits
free-form JSON whose 'segments' list can contain entries like
["CPU Usage", 30] (channel name instead of an int). _seg_from_entry called
int() unconditionally and raised ValueError, killing the whole report run
(model predictor had already finished; result lost).

Fix: wrap _seg_from_entry in try/except (ValueError, TypeError) → return
None for any unparseable entry, so good segments alongside junk are still
extracted. Verified: {["CPU Usage",30],[10,20]} → [(10,20)]. + unit test.
Also archive logs/train_full_run.log (stage1 31250 EMA1.26, stage2 9375 EMA0.75).

tests/test_parse_answer.py

@@ -73,6 +73,18 @@ class TestParseSegments:
         assert any('"a"' in b and '"b"' in b for b in blobs)
         assert '{"d": 4}' in blobs
 
+    def test_non_numeric_segment_entries_are_skipped(self):
+        # a pure-LLM may emit junk like ["CPU Usage", 30] — must not crash, just skip
+        from tsmm.eval.parse_answer import _seg_from_entry
+        assert _seg_from_entry(["CPU Usage", 30]) is None
+        assert _seg_from_entry({"start": "foo", "end": "bar"}) is None
+        assert _seg_from_entry("not a segment") is None
+        # good entries alongside junk are still extracted
+        text = '{"segments": [["CPU Usage", 30], [10, 20]]}'
+        segs, ok = parse_anomaly_segments(text)
+        assert ok is True
+        assert segs == [(10, 20)]
+
     def test_empty_segments_list(self):
         text = '{"segments": [], "type": "none"}'
         segs, ok = parse_anomaly_segments(text)