refactor(events): raise on malformed stream id instead of lex fallback

stream_id_compare's lex-fallback branch was a footgun: a malformed id
that sorts lex-greater than a real one would pin live-tail dedup
forever, dropping every subsequent legitimate event silently. Both
current callers in application/api/events/routes.py pre-validate
inputs against _STREAM_ID_RE before calling, so changing the function
to raise ValueError is a no-op on the happy path and turns the future-
caller footgun into a loud failure.

application/events/keys.py

@@ -35,19 +35,16 @@ def stream_id_compare(a: str, b: str) -> int:
     Stream ids are ``ms-seq`` strings; comparing as strings would be wrong
     once ``ms`` straddles digit-count boundaries. We parse and compare
     as ``(int, int)`` tuples.
+
+    Raises ``ValueError`` on malformed input. Callers must pre-validate
+    against ``_STREAM_ID_RE`` (or equivalent) — a lex fallback here let
+    a malformed id compare lex-greater than a real one and silently pin
+    dedup forever.
     """
     a_ms, _, a_seq = a.partition("-")
     b_ms, _, b_seq = b.partition("-")
-    try:
-        a_tuple = (int(a_ms), int(a_seq) if a_seq else 0)
-        b_tuple = (int(b_ms), int(b_seq) if b_seq else 0)
-    except ValueError:
-        # Malformed id — fall back to lex compare so we at least don't crash.
-        if a < b:
-            return -1
-        if a > b:
-            return 1
-        return 0
+    a_tuple = (int(a_ms), int(a_seq) if a_seq else 0)
+    b_tuple = (int(b_ms), int(b_seq) if b_seq else 0)
     if a_tuple < b_tuple:
         return -1
     if a_tuple > b_tuple:

tests/test_events_substrate.py

@@ -67,11 +67,16 @@ class TestStreamIdCompare:
     def test_missing_seq_treated_as_zero(self):
         assert stream_id_compare("1234", "1234-0") == 0
 
-    def test_malformed_both_sides_falls_back_to_lex(self):
-        # Pure lex compare on garbage — important: this must not raise.
-        assert stream_id_compare("foo", "bar") == 1
-        assert stream_id_compare("foo", "foo") == 0
-        assert stream_id_compare("aaa", "zzz") == -1
+    def test_malformed_input_raises(self):
+        # Callers must pre-validate; the function refuses to silently
+        # lex-compare garbage because a malformed id that sorts
+        # lex-greater would pin dedup forever.
+        with pytest.raises(ValueError):
+            stream_id_compare("foo", "bar")
+        with pytest.raises(ValueError):
+            stream_id_compare("123-0", "foo")
+        with pytest.raises(ValueError):
+            stream_id_compare("foo", "123-0")
 
 
 # ── _format_sse ─────────────────────────────────────────────────────────