Fix handling in-flight requersts in prefetch buffer resize

[knizhnik]

Problem

See #9961
Current implementation of prefetch buffer resize doesn't correctly handle in-flight requests

Summary of changes

1. Fix index of entry we should wait for if new prefetch buffer size is smaller than number of in-flight requests.
2. Correctly set flush position

[github-actions]

7018 tests run: 6710 passed, 0 failed, 308 skipped (full report)

---

Flaky tests (2)

Postgres 16

• test_readonly_node_gc: release-x86-64

Postgres 14

• test_pull_timeline[True]: release-x86-64

Code coverage* (full report)

• functions: 30.4% (8270 of 27227 functions)
• lines: 47.7% (65193 of 136583 lines)

* collected from Rust tests only

---

_{The comment gets automatically updated with the latest test results
cf8809e at 2024-12-02T16:41:15.197Z :recycle:}

[hlinnaka]

Approved as this fixes the flakiness. I'm still pretty scared of the readahead_buffer_resize() function though. It gets executed rarely, and if the compaction logic is wrong, you get subtle, silent corruption.

Calling prefetch_wait_for() in the assign hook is pretty scary too. Most assign hooks are short functions that just do some calculation or transformation and assign the new value to a global variable. But that function does network I/O. I'd suggest just remembering the new value somewhere and applying it later at a convenient time.