test: reduce writeToDisk memory test iterations to fix CI timeout (#92586)
### What?
Reduces `RUNS` from 10,000 to 1,000 in the `next.rs api writeToDisk
multiple times` test in `test/development/basic/next-rs-api.test.ts`.
### Why?
The test was frequently timing out on CI. It spawns a child `node
--expose-gc` process that calls `writeToDisk()` in a loop (`RUNS` times
per batch) for each discovered route (~11 routes), with up to 11
measurement batches per route. At 10,000 runs per batch that is up to
~1.1 million `writeToDisk()` calls total — all under the global
60-second Jest timeout.
Each call, even in the turbo-tasks memoized steady state, has
non-trivial overhead: an NAPI crossing, a task-cache lookup, a
`read_strongly_consistent()` wait, and an O(N effects) iteration over
output assets (each requiring a mutex acquisition). On slow CI hardware
this easily exceeds 60 seconds.
### How?
Reduce `RUNS` to 1,000. This gives a 10× speedup without compromising
the test's ability to detect memory leaks: a leak of even a single OS
page (4 KB) per 1,000 calls is still detectable via the RSS delta check.
The warmup + measurement loop structure is unchanged.
<!-- NEXT_JS_LLM_PR -->
Co-authored-by: Tobias Koppers <sokra@users.noreply.github.com>
Co-authored-by: Claude <noreply@anthropic.com>