sql: improve inner plan leaf transaction tracking

[rharding6373]

This PR changes planner's innerPlansMustUseLeafTxn from being treated as a boolean to a counter instead. As a result, the contract with getFinishedSetupFn has changed so that the returned finishedSetupFn and cleanup functions should only be called once.

This change cleans up how we handle leaf transactions when running a plan inside a plan, e.g., executing UDFs.

Epic: none
Informs: #111097

Release note: None

[cockroach-teamcity]

This change is 

[yuzefovich]

Reviewed 4 of 4 files at r1, all commit messages.
Reviewable status: complete! 0 of 0 LGTMs obtained (waiting on @DrewKimball and @rharding6373)

---

pkg/sql/apply_join.go line 329 at r1 (raw file):

			&subqueryResultMemAcc,
			false, /* skipDistSQLDiagramGeneration */
			atomic.LoadInt32(&params.p.atomic.innerPlansMustUseLeafTxn) == 1,

This now should be >= 1. Perhaps we should introduce a helper method for this?

---

pkg/sql/apply_join.go line 356 at r1 (raw file):

	planCtx := execCfg.DistSQLPlanner.NewPlanningCtx(ctx, evalCtx, &plannerCopy, plannerCopy.txn, distributeType)
	planCtx.stmtType = recv.stmtType
	planCtx.mustUseLeafTxn = atomic.LoadInt32(&params.p.atomic.innerPlansMustUseLeafTxn) == 1

ditto

---

pkg/sql/distsql_running.go line 2028 at r1 (raw file):

		// We already called finishedSetupFn in the previous call to Run, since we
		// only got here if we got a distributed error, not an error during setup.
		dsp.Run(ctx, localPlanCtx, txn, localPhysPlan, recv, evalCtx, nil)

nit: add /* finishedSetupFn */.

---

pkg/sql/planner.go line 184 at r1 (raw file):

	atomic struct {
		// innerPlansMustUseLeafTxn is set to 1 if the "outer" plan is using

nit: update the comment to indicate that it's a counter.

---

pkg/sql/logictest/testdata/logic_test/udf_fk line 490 at r1 (raw file):

statement ok
CREATE TABLE IF NOT EXISTS parent_cascade (p INT PRIMARY KEY);

nit: is this added so that apply_join subtest can be run?

[rharding6373]

TFTR!

Reviewable status: complete! 0 of 0 LGTMs obtained (waiting on @DrewKimball and @yuzefovich)

---

pkg/sql/apply_join.go line 329 at r1 (raw file):

Previously, yuzefovich (Yahor Yuzefovich) wrote…

This now should be >= 1. Perhaps we should introduce a helper method for this?

Thanks for catching this oversight. Done.

---

pkg/sql/apply_join.go line 356 at r1 (raw file):

Previously, yuzefovich (Yahor Yuzefovich) wrote…

ditto

Done.

---

pkg/sql/distsql_running.go line 2028 at r1 (raw file):

Previously, yuzefovich (Yahor Yuzefovich) wrote…

nit: add /* finishedSetupFn */.

Done.

---

pkg/sql/planner.go line 184 at r1 (raw file):

Previously, yuzefovich (Yahor Yuzefovich) wrote…

nit: update the comment to indicate that it's a counter.

Done

---

pkg/sql/logictest/testdata/logic_test/udf_fk line 490 at r1 (raw file):

Previously, yuzefovich (Yahor Yuzefovich) wrote…

nit: is this added so that apply_join subtest can be run?

Yes. Before this change when run alone the subtest would fail.

[yuzefovich]

Reviewed 3 of 3 files at r2, all commit messages.
Reviewable status: complete! 1 of 0 LGTMs obtained (waiting on @DrewKimball and @rharding6373)

---

pkg/sql/apply_join.go line 25 at r2 (raw file):

	"github.com/cockroachdb/errors"
	"github.com/cockroachdb/redact"
	"strconv"

nit: this seems out of order now.

[rharding6373]

Fixing one more formatting issue. The TestStreamingRegionalConstraint failure is from a recently unskipped test (known issue: #112541).

[rharding6373]

TFTR!

bors r+

[craig]

Build succeeded:

• Bazel Essential CI (Cockroach)