colexec: fix recent problem with hash joiner

[yuzefovich]

#53169 has just introduced an optimization in populating of toCheck
slice by having a default hjInitialToCheck slice pre-populated.
However, it could be of insufficient length due to randomization of
coldata.BatchSize() which is now fixed.

Release note: None

[cockroach-teamcity]

This change is 

[asubiotto]

Reviewable status: complete! 1 of 0 LGTMs obtained (waiting on @asubiotto)

[yuzefovich]

TFTR!

bors r+

[asubiotto]

I'm slightly confused though. The error is panic: runtime error: slice bounds out of range [:1588] with capacity 1024. Doesn't that imply that batchSize is larger than it should be?

Reviewable status: complete! 1 of 0 LGTMs obtained (waiting on @asubiotto)

[yuzefovich]

I believe what happens is that init() function is executed before we randomize coldata.BatchSize() value in colexec/main_test.go, so we always create hjInitialToCheck of 1024 length, then we randomize the batch size at which point we might get out of bounds error.

[craig]

Canceled.

[yuzefovich]

I've simply adjusted the comment.

bors r+

[yuzefovich]

bors p=9999

[yuzefovich]

bors r+ p=9999

[craig]

Already running a review

[otan]

i think your series of commands broke bors :P

[craig]

Build failed (retrying...):

• GitHub CI (Cockroach)

[craig]

Build succeeded:

• GitHub CI (Cockroach)