release-24.3: kvstreamer: fix pathological behavior in InOrder mode

[blathers-crl]

Backport 2/2 commits from #134132 on behalf of @yuzefovich.

/cc @cockroachdb/release

---

kvstreamer: fix TestStreamerVaryingResponseSizes

Previously the test was fooling itself - the regex for KV gRPC calls line was incorrect, so it was never matched, and we ended up with unset counter (which happened to pass the test); this is now fixed.

Release note: None

kvstreamer: fix pathological behavior in InOrder mode

This commit fixes the case of pathological behavior by the streamer in the InOrder mode in some cases. Namely, when ordering needs to be maintained, the streamer needs to prioritize sub-requests that have higher "urgency" to be served (i.e. those that are closer to the head of the line). This "urgency" is represented by the values in singleRangeBatch.positions slice where the smaller the value, the higher the urgency, and the value at the zeroth index is used as the priority for the whole single-range batch. It is assumed that the values in this slice are increasing, but this assumption could previously be violated when multiple ranges were touched (when the original batch fit within a single range, we have a separate fast-path that is unaffected by this bug). This was the case because we used mustPreserveOrder = false when instantiating the batch truncation helper. As a result, all sub-requests within the single-range batch would get reordered according to the start key of each request, and the original order wouldn't be restored by the batch truncation helper. This, in turn, would result in the streamer evaluating the requests with effectively random urgency which would then consume the working budget. In the extreme, we would use up all available budget for random requests, buffer them, and would keep on doing so until we get lucky to get the next head-of-the-line request randomly. This is now fixed by restoring the order of positions by the truncation helper when the streamer is in the InOrder mode. This commit also adds a test-only assertion for ensuring the ascending invariant is maintained.

Here is a concrete example of the behavior. Say, we have two ranges [a - f) and [f - ...) and requests

• 0: Get(c)
• 1: Get(e)
• 2: Get(d)
• 3: Get(f)
• 4: Get(a)
• 5: Get(b)

The batch truncation helper will first order all requests by the start key, so it'll process them in the order 4 - 5 - 0 - 2 - 1 - 3. When truncating to the first range [a - f), it'll populate positions as [4, 5, 0, 2, 1] (request 3 is outside of the range, so it'll stop). This slice is what we would previously include into singleRangeBatch.positions, so we would first evaluate the 4th request, then the 5th, etc. Previously, we would also incorrectly compare singleRangeBatches between each other for "in order" priority.

AFAICT this bug has been present since the introduction of the batch truncation helper in 645c154. The assumption of the InOrder mode was already there, in the comment, but wasn't enforced and was overlooked.

Fixes: #133043.

Release note (bug fix): Previously, when executing queries with index / lookup joins when the ordering needs to be maintained, CockroachDB in some cases could get into a pathological behavior which would lead to increased query latency, possibly by several orders of magnitude. This bug was introduced in 22.2 and is now fixed.

---

Release justification: bug fix.

[blathers-crl]

Thanks for opening a backport.

Please check the backport criteria before merging:

• Backports should only be created for serious
  issues or test-only changes.
• Backports should not break backwards-compatibility.
• Backports should change as little code as possible.
• Backports should not change on-disk formats or node communication protocols.
• Backports should not add new functionality (except as defined
  here).
• Backports must not add, edit, or otherwise modify cluster versions; or add version gates.
• All backports must be reviewed by the owning areas TL. For more information as to how that review should be conducted, please consult the backport
  policy.

If your backport adds new functionality, please ensure that the following additional criteria are satisfied:

• 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. State changes must be further protected such that nodes running old binaries will not be negatively impacted by the new state (with a mixed version test added).
• The PM and TL on the team that owns the changed code have signed off that the change obeys the above rules.
• Your backport must be accompanied by a post to the appropriate Slack
  channel (#db-backports-point-releases or #db-backports-XX-X-release) for awareness and discussion.

Also, please add a brief release justification to the body of your PR to justify this
backport.

[cockroach-teamcity]

This change is