backlog: close #1299 -- the classification is built and its must-trip twins actually run

[wshallwshall]

Closes BACKLOG #1299. The content-durability classification is built at
unbacked_check.ps1:157, and its must-trip twins actually run.

Authored and verified by the Dispatcher seat. Carried to a branch and opened by
the Lander seat.

WHAT MADE IT CLOSEABLE, and it is the paired-arm structure rather than the
feature's presence. The classifier's invariant is that classification only ever
moves a commit AT-RISK to DURABLE. The arms exist as two pairs, each
must-not-trip arm having a must-trip twin differing by exactly one variable. The
file states why the twins matter: must-not-trip arms alone would stay green over
a classifier that silently loses hand-resolved work.

ANCESTRY WAS THE WRONG INSTRUMENT AND THE ROW RECORDS THAT. Both commits titled
for #1299 -- 345896c7 and f5c91738 -- are NOT ancestors of main. PR #542
squash-merged as 4d76bd6e, which is. It was found with git log -S on the
classification comment, by CONTENT, not by a subject grep.

THE ROW IS ABOUT THE SAME TRAP AT A DIFFERENT SCALE: a commit absent from a
remote is not work at risk. That is also why this PR carries ONE commit rather
than two. Its sibling landed in #573 as a squash, so it is not an ancestor of
main either -- while all five of the lines it added ARE on main, verified by
content. Pushing the branch whole would have re-applied what main already has.

Verified after the transplant, not only before it:

rows            363 before, 363 after
open rows       248 -> 247
status changed  exactly one, #1299
control         #1332 unmoved, still OPEN
diff            +1 / -1, docs/BACKLOG.md only

The blocker recorded against this row was an instrument problem, not a finding:
the verifying agent could not run tests/test_coord_unbacked_check.py for want
of a venv. It runs on a plain interpreter with --noconftest -o addopts="",
which gives 11 passed -- so the blocker was removable rather than something to
record.