pageserver: remove L0 flush upload wait

[erikgrinaker]

Problem

Previously, L0 flushes would wait for uploads, as a simple form of backpressure. However, this prevented flush pipelining and upload parallelism. It has since been disabled in production and replaced by L0 compaction backpressure.

Touches https://github.com/neondatabase/cloud/issues/24664.

Summary of changes

This patch removes L0 flush upload waits, but still accepts the l0_flush_wait_upload config option as a noop for backwards compatibility with current configurations. The config option will be removed once this has been deployed across the fleet.

This won't be deployed until next week, at which point we should be reasonably sure we won't need to re-enable it.

[github-actions]

7909 tests run: 7523 passed, 1 failed, 385 skipped (full report)

---

Failures on Postgres 14

• test_remote_storage_upload_queue_retries: release-x86-64-with-lfc

# Run all failed tests locally:
scripts/pytest -vv -n $(nproc) -k "test_remote_storage_upload_queue_retries[release-pg14]"

Flaky tests (4)

Postgres 17

• test_remote_storage_upload_queue_retries: release-x86-64-without-lfc

Postgres 16

• test_compute_pageserver_connection_stress: release-x86-64-with-lfc
• test_multiple_subscription_branching: release-arm64-with-lfc

Postgres 15

• test_remote_storage_upload_queue_retries: release-x86-64-with-lfc

Test coverage report is not available

_{The comment gets automatically updated with the latest test results
d10cd01 at 2025-03-12T22:00:53.652Z :recycle:}

[skyzh]

A few tests need to be fixed, otherwise LGTM.