backupccl: replace restore2TB and restoretpccInc tests

[msbutler]

This patch removes the restore2TB* roachtests which ran a 2TB bank restore to
benchmark restore performance across a few hardware configurations. This patch
also replaces the restoreTPCCInc/nodes=10 test which tested our ability to
handle a backup with a long chain.

This patch also adds:

1. restore/tpce/400GB/aws/nodes=4/cpus=16 to measure how per-node throughput
   scales when the per node vcpu count doubles relative to default.
2. restore/tpce/400GB/aws/nodes=8/cpus=8 to measure how per-node throughput
   scales when the number of nodes doubles relative to default.
3. restore/tpce/400GB/aws/backupsIncluded=48/nodes=4/cpus=8 to measure
   restore reliability and performance on 48 length long backup chain relative to
   default.

A future patch will update the fixtures used in the restore node shutdown
scripts, and add more perf based tests.

Fixes #92699

Release note: None

[cockroach-teamcity]

This change is 

[lidorcarmel]

overall makes sense, a few things:

• you're also dropping a tpcc roachperf, right? probably intentional, mention it in the pr/commit description?
• in roachperf we have 2 bank tests with 6 nodes, one has no data, I'm guessing it was removed from restore.go but not from roachperf? or something like that.. which brings me to the question - do you need to remove those from roachperf also?
• the 32 nodes bank test [1] shows a regression since last Dec, we're going to lose this info now. The newly added tests here don't backfill, so maybe we should keep those tests for a few more months to have an overlap with the new tests? WDYT?
  [1] https://roachperf.crdb.dev/?filter=restore&view=restore2TB%2Fnodes%3D32&tab=gce

[msbutler]

you're also dropping a tpcc roachperf, right?

• there's isn't actually a tpcc restore workload in roachperf, but your comment did made me realize I am dropping our current "long chain" roachtest. I will add a new variant of the long chain test once roachtest: simplify name resolution in new restore roachtests #98087 merges.

do you need to remove those from roachperf also?

• i'm assuming that the DR team will stop using roachperf in a month or so once we can observe test perf in prom/grafana. So I dont think we need to do any extra clean up in roachperf. I'd rather let it rot >:)

to paraphrase your last question: " should we keep them around for a couple months for perf regression detection?"

• the roachperf graphs for the old tests aren't getting deleted, nor do I think anyone is actively looking into that regression. - Further these bank2tb tests aren't a very good signal of restore perf as the fixtures were created in 2019. IMHO I don't think we can learn much wrt restore perf from these old tests. I also think the new tests provide much more general test coverage.
• We also need to delete them in the next few weeks anyway to let backupccl: disallow restore of backups older than the minimum supported binary version #93804 pass.

[lidorcarmel]

the roachperf graphs for the old tests aren't getting deleted, nor do I think anyone is actively looking into that regression. - Further these bank2tb tests aren't a very good signal of restore perf as the fixtures were created in 2019

I think it is helpful to have it, it helps us understand that things regressed slowly in that time range. I know it's not the best test but it's info I'd like to have available for debugging. anyway, sounds like we'll have the history so we're good, thank you!

[msbutler]

TFTR!

bors r=lidorcarmel

[craig]

Build succeeded:

• Bazel Essential CI (Cockroach)