comments: drop retired lint markers, re-cite migrated carriers

peel-ok/sizelint-ok/primsize-ok annotations lose their tools; sites
keep the WHY in plain words. Citations of retired carriers move to
their fixture or @test successors (949_errtype_compare -> r949_*,
900_stdlib -> library owners).

cmd/w6c/cgen.c

@@ -770,10 +770,9 @@ nullable_ptr_tag(Type *t)
 	if (t == NULL || t->kind != TY_TAGGED) return 0;
 	int i = 0;
 	for (Tparam *p = t->params; p; p = p->next, i++) {
-		/* peel-ok: single peel PROBE-CLEARED (batch-2 c3-B2,
-		 * 018ef66) — constructible variant params never carry
-		 * 2+-level NAMED at this scan; ww twin cgenutil.ww:2758
-		 * carries the identical annotated peel. */
+		/* A single-level inspection is sufficient here (batch-2 c3-B2,
+		 * 018ef66): constructible variant params never carry nested
+		 * NAMED layers at this scan; the ww twin uses the same invariant. */
 		Type *pu = (p->type && p->type->kind == TY_NAMED)
 		    ? p->type->under : p->type;
 		if (pu && pu->kind == TY_PTR) return i;
@@ -849,7 +848,7 @@ cg_variant_match(Type *vt, Type *src)
 		 * Sound only while the model is nominal-lossy; the collision
 		 * guard at the widen site (cg_widen_tagged_store) enforces the
 		 * invariant for when #199b/B-full lands true nominal layout.
-		 * peel-ok (#218, B5-c1): these single peels serve ONLY the
+		 * These single-level inspections serve ONLY the
 		 * both-TAGGED structural fallback — a NAMED struct source vs
 		 * a NAMED variant falls through every arm at ANY depth
 		 * (.ai/ken-b5-oracle.md §4: kb5_v2s1i both-wrong-identical
@@ -875,7 +874,7 @@ cg_variant_match(Type *vt, Type *src)
 static int
 cg_variant_struct_match(Type *vt, Type *src)
 {
-	/* peel-ok (#218, B5-c1): shape-only collision count for the
+	/* Shape-only collision count for the
 	 * both-TAGGED fallback above — same probe record, task #95
 	 * (.ai/ken-b5-oracle.md §4). */
 	Type *vu = (vt && vt->kind == TY_NAMED) ? vt->under : vt;