sql: add safety gates for pausable portal cleanup and flow invalidation

[ZhouXing19]

Informs: https://github.com/cockroachlabs/support/issues/3463

When a pausable portal encounters an error during execution, two issues can lead to panics on subsequent resume attempts:

1. The underlying FlowBase gets reset to nil during cleanup, but the portal's flow reference remains non-nil, causing hasFlowForPausablePortal() to incorrectly return true.

2. Errored portals are not removed from the portal map because deletion only occurs when execStmt() returns a non-nil fsm.Event.

This change adds two defensive checks:

• Nil the whole flow object hanging off the portalInfo while cleaning up the flow.
• Ensure errored portals are properly cleaned up regardless of event state

These gates prevent nil pointer dereferences when resuming portals that have been partially cleaned up due to errors.

Release note: None

[blathers-crl]

It looks like your PR touches production code but doesn't add or edit any test code. Did you consider adding tests to your PR?

_{🦉 Hoot! I am a Blathers, a bot for CockroachDB. My owner is dev-inf.}

[cockroach-teamcity]

This change is 

[yuzefovich]

@yuzefovich reviewed 4 of 4 files at r1, all commit messages.
Reviewable status: complete! 0 of 0 LGTMs obtained (waiting on @ZhouXing19)

---

-- commits line 23 at r1:
nit: I'd probably omit a release note here given that we cannot explain well to users which conditions were necessary for the bug to happen.

---

pkg/sql/conn_executor_exec.go line 1201 at r1 (raw file):

			updateRetErrAndPayload(retErr, retPayload)
			portal.pauseInfo.resumableFlow.cleanup.run(ctx)
			portal.pauseInfo.resumableFlow.flow = nil

nit: rather than unsetting flow after each time cleanup runs, we should modify the cleanup function itself, this will be more bullet-proof.

[ZhouXing19]

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

---

-- commits line 23 at r1:

Previously, yuzefovich (Yahor Yuzefovich) wrote…

nit: I'd probably omit a release note here given that we cannot explain well to users which conditions were necessary for the bug to happen.

Done.

---

pkg/sql/conn_executor_exec.go line 1201 at r1 (raw file):

Previously, yuzefovich (Yahor Yuzefovich) wrote…

nit: rather than unsetting flow after each time cleanup runs, we should modify the cleanup function itself, this will be more bullet-proof.

Good idea! Done.

[yuzefovich]

Nice! I think it's worth backporting to 25.3 and 25.4, so I added the labels. Also make sure to update the PR description.

@yuzefovich reviewed 3 of 3 files at r2, all commit messages.
Reviewable status: complete! 1 of 0 LGTMs obtained (waiting on @ZhouXing19)