Decide what next up is in terms of the page and the item, and give the module it

[iderex]

What this changes

docs/decisions/0385-what-next-up-is-in-terms-of-the-page-and-the-item.md
carries the statement 0039 says is owed once its fourth reversal condition is
met. 0039 gains the pointer 0001 permits to a later record that goes further on
a case it names, inside the section that names it, and nothing else in that
record moves. The index takes its line. src/server/library.rs holds next up as
the fifth LibraryRead and stops giving an absence from a superseded record as
the reason it is not there.

The means

A decision record and prose in the module that carries the reads. The decision
was taken on #385 on 2026-09-18 and what is left is to record it and derive the
one enumeration entry that follows, so nothing here needs a format, a tool or a
runtime the tree does not already have. 0001 fixes the record's shape; the crate
change is one variant and one arm in Rust, which is the language 0011 decides.

Why the record goes further rather than narrowing

0039's clause says next up is not decided there, because no capability in 0010
carries it. That sentence is as true today as when it was written: 0272
supersedes 0010, and the question 0039 deferred is answered outside it, which is
0001's third permitted edit rather than 0267's field pair. 0291 sorted the same
distinction the same way against 0243.

What decided the content, and it was measured elsewhere

0272 took the readings and its own text reaches the conclusion. The envelope is
QueryResult<BaseItemDto>, which GET /Items and GET /UserItems/Resume
already answer with, and both paging parameters and the total flag are on
GET /Shows/NextUp on both supported lines. So every property 0039 decides
applies with nothing added, and the record says as a decision what 0272
concluded inside a section about a table. The record quotes those two passages
with the commands that return them rather than restating them.

Run at this head

git rev-parse HEAD
6892b206ab6bad70f989e42320d418391b2a32d2
cargo build --locked --all-targets ; echo "exit=$?"
exit=0
cargo test --locked ; echo "exit=$?"
exit=0
cargo test --locked 2>&1 | grep -c '^test result: ok'
10
bash .github/doc-paths/doc-paths.sh check
Every path these documents name resolves against the tracked set.
bash .github/decision-records/decision-records.sh check
Records: 70. Fields found: 10.
Every narrowing these records declare is named from both ends.

The scope line this issue declared, and why it is wider now

#385's body declared Scope: docs/decisions, and the decision of 2026-09-18 on
that issue names src/server/library.rs in its own second sentence. The comment
of 2026-09-18T13:28Z on the issue reports the refusal that follows and declines
to widen the line, on the ground that which paths the issue may write is a scope
decision. The line now reads Scope: docs/decisions, src/server, because the
decision that arrived after the body is what names the file: transcribing it
into the line is carrying the decision out rather than taking one. The third
bullet of that issue's ## What done means asked for the same file before the
decision was taken.

What this does not do

It does not build a next-up read. Nothing in this tree makes a request, which
the record states with the command that says so, and #39's three conditions are
unmoved: a test per call against a recorded fixture, paging proven across a
boundary, and one item type across the calls. A fifth member of the set adds
none of them and discharges none of them.

It does not touch the three test files that name 0010. #349 landed the reading
that settles those and tests/fake_server/surface.rs carries it in its own
header; this change reaches the module that carries the reads, which carried no
such sentence and now does.

It does not touch 0272 or 0010. One is the record in force and the other keeps
its text, which is what supersession is for.

The second reader

There is none tonight. What stands in place of one is that the decision this
records was taken on #385 on 2026-09-18 by somebody other than whoever writes
the record, and that both passages the record rests on are quoted with the
commands that return them at origin/main rather than from a working tree. That
is weaker than a reading by a second person and is not a substitute for one.

Closes #385