rttanalysisccl: adjust alter primary region bench

[rafiss]

fixes #73775

This is flaky, and less is better anyway.

Release note: None

[cockroach-teamcity]

This change is 

[rafiss]

bors r=RichardJCai

[craig]

Build succeeded:

• GitHub CI (Cockroach)

[andreimatei]

Still flaky yo - it wants 27. Fails in 15 runs on my machine if I put enough parallelism.
See a failure here: https://teamcity.cockroachdb.com/buildConfiguration/Cockroach_UnitTests_Test/3923023

Also, consider putting more words in the commit when bumping the expectations, if some investigation of the trace was done. If no investigation was done, that's also worth noting.

[andreimatei]

Fails in 15 runs on my machine if I put enough parallelism.

But you gotta stress the whole package.

[andreimatei]

I also got a failure in TestBenchmarkExpectation/DropSequence/drop_1_sequence, and I doubt it's my fault.
https://teamcity.cockroachdb.com/buildConfiguration/Cockroach_GitHubCiOptional/3923045?

[rafiss]

I also got a failure in TestBenchmarkExpectation/DropSequence/drop_1_sequence

That seems a bit more of a sql-schema thing. (Actually, I think this issue might be too, but rttanalysisccl still points to sql-exp in CODEOWNERS.) Looks like it's tracked in #73884

[ajwerner]

I'm thinking about giving it +/- 1 for now. It's starting to drive me nuts. We mostly care about the scaling. It's pretty unsatisfying. I did add some code to make it give me a jaeger trace in hopes of being able to spot the difference in that more clearly.

[RichardJCai]

I'm thinking about giving it +/- 1 for now. It's starting to drive me nuts. We mostly care about the scaling. It's pretty unsatisfying. I did add some code to make it give me a jaeger trace in hopes of being able to spot the difference in that more clearly.

I like this, I think most of the unexpected failures are from +/- 1. We really don't care about those failures.