zeph: docstring pins from reply 146 (token-shaped mutants; value dedup, not prefix+len)

wizard-work/zeph/q60_enforce.py

@@ -91,6 +91,10 @@ import sys
 import tempfile
 
 TRUNC_RE = re.compile("([0-9a-f]{8,}(?:-[0-9a-f]{4})*)-\u2026")
+# Mutant note (reply 146 section 1): an injected token must be
+# token-shaped -- stem + hyphen + U+2026 ellipsis. A bare stem, or a stem
+# with three ASCII dots, never matches TRUNC_RE and leaves --selftest
+# green, which reads exactly like a mutant that failed to bite.
 FULL_RE = re.compile(
     r"[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}"
 )
@@ -264,6 +268,11 @@ def run_check(tree, excludes):
     # fault re-commits the fault at the next head. Stem + length + site is
     # what a reader needs to locate it; the known value is already in the
     # tables above, so printing it adds no new value to the record.
+    # Masking note (reply 146 section 3): set() dedupes by VALUE, not by
+    # (prefix, length), so two distinct values sharing prefix and length
+    # print two identical masked lines -- multiplicity stays visible, and
+    # a change that deduped by (prefix, len) "to cut noise" would delete
+    # the last evidence of such a pair. Do not make that change.
     for rel, text in contents.items():
         for val in set(FULL_RE.findall(text)):
             prefix = val[:8]