Thirty releases, D-243 ... D-273, waves W13 through W16, against a written
review of 0.15.0: twenty-four numbered code findings and six architectural
notes, twenty-seven of thirty closed.

Schema v15 -> v18 over three rungs. Snapshot container v4 and its v3 header
unchanged. DbError 41 -> 42 variants. Public surface 1,624 -> 1,763 items.
MSRV 1.88 unchanged.

The shape of the cycle is that the review was reliably right that something
was wrong and much less reliably right about why, and every correction here
is a measurement rather than a re-reading. Its only High was High. Its C-4
asked for an index that is never faster than no index. Its C-11 named three
structs where the defect was twenty-one structs wide. Its C-8's suggested fix
repeats the bug it describes one line further down the query. Its C-13, filed
Low with the words "nothing downstream is wrong", was a live data-
availability bug that made an ordinary public call render a database's whole
history unreadable.

And the largest single defect of the cycle is not in the review at all:
links_at_tx had been running as a co-routine inside the recursive step since
0.13.2 -- 10.6 s where the answer takes 59 ms -- hidden from eleven releases
of branching probes because they were all shaped to time the branched read,
and the branched read's window materialises on its own.

Of the plan's ten acceptance criteria, seven met, two restated then met, and
the tenth met late, after two further releases, because reaching it meant
first building the thing that could check it.

Three items remain open by decision and are named in the release note: A-4
(the archive module split, which this cycle made worse rather than better),
A-5, and cargo-semver-checks, which is at 1.0 because it wants a released
baseline to diff against. This tag is that baseline.

Verified for this commit, locally:
  757 Rust across 59 targets - 736 with --no-default-features across 58 -
  607 Python passed / 2 skipped - clippy clean under -D warnings with metrics
  and property-tests - rustdoc clean - fuzz crate compiles - public API gate
  at 1,763 items, unchanged.

CI is green on this SHA across all eight jobs, read rather than glanced at
(D-234), run 34162891774: the suite on three runners, the feature-off suite
run on ubuntu rather than compiled, clippy, MSRV, the fuzz crate, the public
API baseline, and cargo publish --dry-run. The quarantined property arm is
non-blocking by D-236 and returned crashed-R15 on Windows with zero named
failures, which is D-147's documented Windows fault.

See docs/releases/v0.16.0.md.