Read 0291's fourth reversal condition as a price paid rather than as a trigger

[iderex]

What changed

docs/decisions/0351-the-condition-that-counts-this-board.md is added. It reads
0291's fourth reversal condition as a price this board has already paid rather
than as a trigger that can fire, and says why the count that condition names has
a floor of one. 0291 gains the Narrowed-by: field 0267 requires in the other
direction, and docs/decisions/README.md gains the index line.

It goes further than 0291 rather than editing it. Rewording what would reverse a
landed record is not one of the edits 0001 allows, and #351's own done-condition
asks for a record for that reason.

The specific failure it prevents

A reader of 0291 meets four reversal conditions and has no way to tell that one
of them cannot happen. Taking it for a trigger means waiting for a state this
board does not produce, which is the failure the condition was written against
arriving through the condition itself. Noticing that it cannot fire means
re-deriving the chain behind #98, the label history and the reason of
2026-08-22 from the tracker, which is an hour of reading that ends where the
record now is.

Evidence

Read on 2026-09-18 at the mainline this branch is cut from,
243f6e1.

The condition, as it stands in the record:

git show origin/main:docs/decisions/0291-the-provider-the-socket-waits-on.md | sed -n '226,232p'
The count of open issues on this board that can be worked without a request
reaches zero. The wait is then the whole of this board's throughput rather than
one milestone's, which is a different price from the one weighed above, and the
choice is retaken against it:

    gh issue list --repo Flowfin/core --state open --limit 500 --json number,labels \
      --jq '[.[] | select([.labels[].name] | map(startswith("blocked-on-")) | any | not)] | length'

What that command answers, and who is in the set:

gh issue list --repo Flowfin/core --state open --limit 500 --json number,labels \
  --jq '[.[] | select([.labels[].name] | map(startswith("blocked-on-")) | any | not)] | length'
8
gh issue list --repo Flowfin/core --state open --limit 500 --json number,title,labels \
  --jq '[.[] | select([.labels[].name] | map(startswith("blocked-on-")) | any | not)] | sort_by(.number) | .[] | "#\(.number) [\([.labels[].name]|join(","))] \(.title)"'
#86 [security,ci,tests] Fuzz the parsers, and replay the seed corpus inside the gating build
#88 [ci,tests] Build and test against the oldest server surface the core claims to support
#94 [ci,release] Build and publish the release artifacts, reproducibly
#98 [documentation,planning] The plan for this repository, in build order, and what it leaves out
#104 [tests] Keep the recorded fixtures honest against a real server
#291 [ci,decision] Decide how the target leg compiles the crypto provider 0243 brings, or which triples leave 0113
#351 [documentation,decision] 0291's fourth reversal condition has one member left, and no state of this board takes it to zero
#385 [documentation,decision] 0039's fourth reversal condition is met by 0272, and the module that carries the reads still gives the absence it rested on

#98's first exit, closing, is downstream of the wait:

gh issue view 98 --repo Flowfin/core --json body --jq .body | grep -A1 'The first release in M11'
The first release in M11 exists and an operator has run the probe against their
own server. Until then this issue stays open as the entry point to the board.

for n in 92 39 27 291; do printf '%s\t' "$n"; gh api "repos/Flowfin/core/issues/$n/comments?per_page=100" --paginate --jq '[.[] | .body | split("\n")[0] | select(startswith("Waiting for:"))] | last'; done
92      Waiting for: Flowfin/core#39
39      Waiting for: Flowfin/core#27
27      Waiting for: Flowfin/core#291
291     Waiting for: act - a pure-Rust crypto provider for `rustls` reaching a release the target leg compiles on every triple in 0113's set (upstream). Unchanged, and re-read below rather than carried over.

Its second exit, a blocked-on- label, was applied and removed twice and argued
shut on the issue itself:

gh api repos/Flowfin/core/issues/98/timeline --paginate --jq '.[] | select(.event=="unlabeled" or .event=="labeled") | "\(.created_at) \(.event) \(.label.name)"'
2026-08-07T22:46:41Z labeled documentation
2026-08-07T22:46:41Z labeled planning
2026-08-17T13:47:25Z labeled blocked-on-issue
2026-08-18T21:09:59Z unlabeled blocked-on-issue
2026-08-22T08:38:19Z labeled blocked-on-issue
2026-08-22T16:14:57Z unlabeled blocked-on-issue

The comment of 2026-08-22T16:14:55Z gives the reason, and the record quotes it
rather than restating it.

The correction this change carries

#351's body derives the count as one with #98 as its only member, and the
decision of 2026-09-18 on that issue restates that figure. The same command
answers eight today, as pasted above. The reading is unchanged by it, because
what the argument needs is the floor and not the number - both of #98's exits
are shut at any count, and eight is no closer to zero than one - but the
difference is real and is written into the record under its own heading rather
than left for a reader to trip over. It was found by running the command before
writing the record rather than by carrying the figure out of the issue.

The guards this ran through

No guard is added or edited, so there is no deliberate violation to show. The
checks that read this register were run at the commit being pushed:

bash .github/decision-records/decision-records.sh check
Records: 68. Fields found: 10.
ok    10 field(s) name a record that exists, name a clause, and are named back

bash .github/doc-paths/doc-paths.sh check
ok    82 document(s), 411 named path(s) resolved, 0 excused as examples

bash .github/invariants/invariants.sh check
Every rule above was applied to its subject and refused nothing.

bash .github/format/format.sh check
Every tracked source file above is written the way the formatter would write it.

The Narrowed-by: field on 0291 is what takes the field count from 8 to 10; it
and the new record's Narrows: line are one pair, and the check reads it in both
directions.

Three files are touched, all inside the docs/decisions scope #351 declares:

git diff --name-only origin/main...HEAD
docs/decisions/0291-the-provider-the-socket-waits-on.md
docs/decisions/0351-the-condition-that-counts-this-board.md
docs/decisions/README.md

No cargo run was made, because nothing this change touches is compiled.

What this does not cover

• The choice 0291 holds. The decision, its costs and the three conditions
  that do fire are untouched. This reads the fourth and takes no position on
  whether waiting is still right.
• The wording of the condition. It stands in 0291 exactly as written. What
  is added is how it is read, which is the form 0291's fourth reversal condition has one member left, and no state of this board takes it to zero #351's done-condition asks for
  and the only one 0001 allows.
• Whether The plan for this repository, in build order, and what it leaves out #98 should ever be closed or labelled. Both exits are read as they
  stand. Neither is argued for or against here.
• Anything that fires the other three conditions. A provider reaching a
  release, a runner image measured with a working C cross-toolchain, and a triple
  leaving 0113's set are states of the world outside this tracker, and none of
  them is read here.
• The count's definition. What subtracts an issue is a blocked-on- label
  and nothing else. That rule is neither widened nor narrowed.
• Decide how the target leg compiles the crypto provider 0243 brings, or which triples leave 0113 #291 itself. Its done-condition is about a socket, a record and a green
  target leg, and nothing here moves it.

Who has read this

Nobody other than its author. There is no second reader on this board tonight,
and every reading above is quoted with the command that produced it so it can be
repeated rather than trusted.

Closes #351