fix(pageserver): fix gc-compaction racing with legacy gc

[skyzh]

Problem

close #10049, close #10030, close #8861

part of #9114

The legacy gc process calls get_latest_gc_cutoff, which uses a Rcu different than the gc_info struct. In the gc_compaction_smoke test case, the "latest" cutoff could be lower than the gc_info struct, causing gc-compaction to collect data that could be accessed by latest_gc_cutoff. Technically speaking, there's nothing wrong with gc-compaction using gc_info without considering latest_gc_cutoff, because gc_info is the source of truth. But anyways, let's fix it.

Summary of changes

• gc-compaction uses latest_gc_cutoff instead of gc_info to determine the gc horizon.
• if a gc-compaction is scheduled via tenant compaction iteration, it will take the gc_block lock to avoid racing with functionalities like detach ancestor (if it's triggered via manual compaction API without scheduling, then it won't take the lock)

[github-actions]

7077 tests run: 6762 passed, 0 failed, 315 skipped (full report)

---

Flaky tests (5)

Postgres 17

• test_timeline_archive[4]: debug-x86-64

Postgres 16

• test_prefetch[4]: release-x86-64
• test_prefetch[None]: release-x86-64

Postgres 15

• test_pull_timeline[True]: release-x86-64

Postgres 14

• test_prefetch[4]: release-arm64

Code coverage* (full report)

• functions: 31.4% (8331 of 26528 functions)
• lines: 47.7% (65613 of 137438 lines)

* collected from Rust tests only

---

_{The comment gets automatically updated with the latest test results
9ddd59d at 2024-12-09T20:16:43.711Z :recycle:}