release-21.1: kvserver: consider suspect stores "live" for computing quorum

[aayushshah15]

Backport 1/1 commits from #67714.

/cc @cockroachdb/release

---

Previously, when making the determination of whether a range could achieve
quorum, the allocator ignored "suspect" stores. In other words, a range with 3
replicas would be considered unavailable for rebalancing decisions if it had 2
or more replicas on stores that are marked suspect.

This meant that if a given cluster had multiple nodes missing their liveness
heartbeats intermittently, operations like node decommissioning would never
make progress past a certain point (the replicate queue would never decide to
move replicas away because it would think their ranges are unavailable, even
though they're really not).

This patch fixes this by slightly altering the state transitions for how stores
go in and out of "suspect" and by having the replica rebalancing code
specifically ask for suspect stores to be included in the set of "live"
replicas when it makes the determination of whether a given range can achieve
quorum.

Release note (bug fix): A bug that was introduced in 21.1.5, which prevented
nodes from decommissioning in a cluster if it had multiple nodes intermittently
missing their liveness heartbeats has been fixed.

[blathers-crl]

Thanks for opening a backport.

Please check the backport criteria before merging:

• Patches should only be created for serious issues.
• Patches should not break backwards-compatibility.
• Patches should change as little code as possible.
• Patches should not change on-disk formats or node communication protocols.
• Patches should not add new functionality.

If some of the basic criteria cannot be satisfied, ensure that the exceptional criteria are satisfied within.

• There is a high priority need for the functionality that cannot wait until the next release and is difficult to address in another way.
• The new functionality is additive-only and only runs for clusters which have specifically “opted in” to it (e.g. by a cluster setting).
• New code is protected by a conditional check that is trivial to verify and ensures that it only runs for opt-in clusters.
• The PM and TL on the team that owns the changed code have signed off that the change obeys the above rules.

Add a brief release justification to the body of your PR to justify this backport.

Some other things to consider:

• What did we do to ensure that a user that doesn’t know & care about this backport, has no idea that it happened?
• Will this work in a cluster of mixed patch versions? Did we test that?
• If a user upgrades a patch version, uses this feature, and then downgrades, what happens?

[cockroach-teamcity]

This change is 

[nvanbenschoten]

Reviewable status: complete! 1 of 0 LGTMs obtained

[nvanbenschoten]

However, I have some concern that we haven't dug into recent roachtest failures like #68171 and #68169 enough to consider #67714 properly baked. And the timing on the roachtest flakiness is, for lack of a better word, suspect.

[aayushshah15]

Yeah, those do seem like they could be related. It should be fine to wait on merging this backport for a bit longer.

[aayushshah15]

I have some concern that we haven't dug into recent roachtest failures like #68171 and #68169

Those turned out to be unrelated, so I'll merge this on green to get it into 21.1.8.