release-21.2: kvcoord: avoid concurrent rollbacks when making parallel commits explicit

[blathers-crl]

Backport 1/1 commits from #69966 on behalf of @erikgrinaker.

/cc @cockroachdb/release

---

TxnCoordSender allows EndTxn(commit=false) rollback requests even if
the transaction state is finalized, since clients can send multiple
rollbacks (e.g. due to context cancellation). However, it allowed this
even when the transaction was committed. This could pass the request
through while the txnCommitter was asynchronously making an implicit
commit explicit, which would violate the txnLockGatekeeper requirement
that transaction requests are synchronous (non-concurrent) which would
return an unexpected error for the rollback.

This patch rejects additional EndTxn(commit=false) requests if the
finalized transaction is known to be committed, to prevent this race
condition. If rejected, the returned error is of the same type that
would be returned by EndTxn evaluation, although with a different
message string.

Note that even though the returned error should really have
REASON_TXN_COMMITTED in this case, which is also what txn.Rollback()
expects in order to omit logging, the current EndTxn code incorrectly
returns REASON_TXN_UNKNOWN in this case. This behavior is retained to
minimize the change, but should be corrected separately.

Resolves #68643.
Informs #69965.

Release justification: fixes for high-priority or high-severity bugs in existing functionality
Release note: None

---

Release justification:

[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.
• Patches must not add, edit, or otherwise modify cluster versions; or add version gates.

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 

[erikgrinaker]

Merge already approved in latest release meeting, so merging despite downgrade to GA-blocker.