docs: describe the readiness gate that exists, not the one that was deleted

[Yaraslaut]

What this is

Five header comments cited tryFireImpl and SubscriberEntry. Neither name exists anywhere in the tree — git grep -w finds them only in the comments citing them.

The archaeology

The handler-side draft machinery tryFireImpl belonged to was deleted. Today BridgeHandler has no set<> (only setDefaultSession / setExecuteDeadline / setPrincipal), no draft, no entries, and no SubscriberState.

But the gate was not deleted — it moved. The readiness check tryFireImpl performed now runs in morph::flows::FlowSession::set<>:

// include/morph/forms/flows.hpp:267
if (::morph::model::ActionValidator<A>::ready(draft)) {
    fireStep<A>(std::move(draft), stepIndex);
}

and the comment five lines above it says so in as many words: "The readiness gate that used to live in the handler's draft machinery lives here now."

So these citations are repointed rather than deleted, which keeps the explanation each was carrying. ValidationError's "two execution sites reached without passing a client-side gate" arithmetic in registry.hpp is still exactly true — it was only naming the wrong gate.

The five sites

Site	Change
core/bridge.hpp:2317	reactive counterpart of the request/reply validator gate -> FlowSession::set<>
core/registry.hpp:126	Doxygen, published API — ValidationError's gate pair, renamed
core/registry.hpp:413	same gate pair, server dispatch path
forms/flows.hpp:189	dead SubscriberEntry + a false attribution (below)
tests/test_coverage_gaps.cpp:575	dead section marker removed

Two sites needed no successor name:

• flows.hpp:189 attributed "exactly one SubscriberEntry per action type" to docs/spec/core/bridge.md's "Subscription semantics". That section says the opposite axis — "subscribe<R>(cb) is keyed on the result/state type, not on an action" — and its cardinality rule is "one callback per (handler, R)". The false attribution goes; the real constraint behind the AllDistinct<Steps...> assert is named instead: one slot of FlowSession's own std::tuple<Steps...> _drafts.
• tests/test_coverage_gaps.cpp:575 was a bare section marker with no test under it, for a coverage gap in a deleted function. Its cited range bridge.hpp:498-499 is now mid-doc-comment on attachHandlerAsync, so the line numbers were stale too. Removed.

Verification

Comment-only: every changed line is a comment (git diff -U0 filtered for non-comment lines returns nothing).

• scripts/check_spec_citations.sh — exit 0, "Prose lint OK: every pinned fact is still cited; no banned terminology found; every cited path resolves; every cited section exists" (720 references, 60 cited sections)
• scripts/check_deprecated_markers.sh, check_test_type_names.sh, check_journal_stamps.sh — exit 0
• Full clang-debug build under -Weverything (Clang 22.1.8, 29 flags, strict) — clean
• ctest — full suite

Refs #355

🤖 Generated with Claude Code

https://claude.ai/code/session_0154xzWuBMPveLcdeUgydifb

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!