kvserver: remove above-raft throttling of AddSST

[tbg]

We now rely on admission control to protect the nodes.¹

Epic: none
Release note (ops change): The following cluster settings were retired and no longer have any effect:

• kv.bulk_io_write.concurrent_addsstable_requests
• kv.bulk_io_write.concurrent_addsstable_as_writes_requests
• rocksdb.ingest_backpressure.l0_file_count_threshold
• rocksdb.ingest_backpressure.max_delay

Footnotes

1. https://github.com/cockroachdb/cockroach/pull/98762#discussion_r1164701947 ↩

[blathers-crl]

It looks like your PR touches production code but doesn't add or edit any test code. Did you consider adding tests to your PR?

_{🦉 Hoot! I am a Blathers, a bot for CockroachDB. My owner is dev-inf.}

[cockroach-teamcity]

This change is 

[tbg]

@irfansharif could you give me an explanation on why we're now fine removing this? Because we have sufficient integration with admission control? Or because we just don't think this will do more harm than good?

[tbg]

Friendly ping.

[irfansharif]

I miscommunicated the sequencing, I don't think we should get rid of these for at least a few weeks. Not until #98308 has landed and has had chance to bake on master as enabled. This is just out of an abundance of caution and not inviting more CI fallout than necessary. What I can do instead is assign myself this PR, mark it as draft, and take it over the finish line when ready.

[irfansharif]

Because we have sufficient integration with admission control? Or because we just don't think this will do more harm than good?

Yes to both, after the kvflowcontrol machinery is used by default, we should have sufficient protection using just AC machinery and I don't think these limiters would really be meaningful.

[tbg]

👍🏽 Would you mind re-opening under your name? That way it doesn't dangle around in my workflows. Thanks!

Filed #102683 to track and assigned to you.