just little things

Reviewed 1 of 1 files at r1, 1 of 1 files at r2, 1 of 1 files at r3, 1 of 1 files at r4, 2 of 2 files at r5, 1 of 1 files at r6, 1 of 1 files at r7, 3 of 3 files at r8.
Reviewable status: complete! 0 of 0 LGTMs obtained (waiting on @ajwerner and @jayshrivastava)

pkg/workload/schemachange/error_screening.go, line 228 at r8 (raw file):

		previousRows := map[string]bool{}

		for _, row := range rows {

this logic could be refactored to be more readable. Consider factoring out a helper per row, constraint tuple. I also wonder if you could rework this to be one query per constraint.

pkg/workload/schemachange/operation_generator.go, line 361 at r4 (raw file):

	}

	ifNotExists := og.randIntn(2) == 0

nit: propagate some configuration of this rate lest we forget that we have control?

pkg/workload/schemachange/operation_generator.go, line 1283 at r4 (raw file):

}

// randTable returns a sequence qualified by a schema

nit: randSequence

pkg/workload/schemachange/operation_generator.go, line 381 at r5 (raw file):

	// Randomly decide if the sequence should be owned by a column. If so, it can
	// set set using the tree.SeqOptOwnedBy sequence option.
	if og.randIntn(2) == 0 {

same nit about control.

pkg/workload/schemachange/operation_generator.go, line 1030 at r6 (raw file):

	}

	srcSequenceExists, err := sequenceExists(tx, srcSequenceName)

I wonder if we're going to want to eventually put a transaction scoped cache on this sort of information. Just food for thought.

Reviewed 3 of 3 files at r17, 1 of 1 files at r18.
Reviewable status: complete! 1 of 0 LGTMs obtained