four nits from the itd-189 fix delta review including an inverted test failure message and an over broad negative loop

four nits from the itd-189 fix delta review including an inverted test failure message and an over broad negative loop

Four nits from the itd-189 fix-delta ruthless review, settled at the ship commit.

1. schema_test.go -- a failure message is inverted and its verb is useless: it fires precisely when the gate does NOT read the value as absent, and prints a constant false inside that branch.
2. schema_test.go -- the negative loop rejects the substrings "skipped" and "refuses" across EVERY record_schema finding on those two paths rather than the duplicate-key one, so an unrelated future message on an admission or a disposition trips it with a misleading reason.
3. commands/capture.md -- the spellings list names "an alias" as carrying nothing. An alias to a defined anchor carries the anchor's value; it passes the gate regardless, so only the "carry nothing" half is the overclaim. Adjacent to but distinct from iss-2608301808197261 item 4.
4. schema_test.go -- a comment calls its list "the whole of what the enumeration claims", but commands/capture.md also lists a block scalar holding nothing, which the pin omits because it is valueEmpty's business rather than isAbsentValue's. The two lists the test claims to bind are not the same set.

1. Added from the fix-delta SECURITY review, which offered it as a nit rather than a finding: commands/capture.md does not name Null/NULL or U+00A0 among the refused spellings, although both ARE refused. It errs in the safe direction -- an author using them gets a clear refusal rather than a silent pass -- and the doc's category framing ("a YAML null", "whitespace") arguably covers them. Folded here rather than given an id.

Grounds

• pursued: an unrelated record_schema message on those paths no longer trips the loop, and every spelling capture.md lists is one the gate refuses; a listed spelling that passes would show it wrong