txnwait: do not require vmodule to log deadlocks in a workload

[arulajmani]

Deadlocks in a workload are rare (and bad) enough that they should be logged at the level of Info, instead of under vmodule.

Epic: none

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 

[nvanbenschoten]

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

---

pkg/kv/kvserver/txnwait/queue.go line 722 at r1 (raw file):

				p1, p2 := pusheePriority, pusherPriority
				if p1 < p2 || (p1 == p2 && bytes.Compare(req.PusheeTxn.ID.GetBytes(), req.PusherTxn.ID.GetBytes()) < 0) {
					log.Infof(

Do you think we should wrap this in an EveryN? Maybe only log it once a second, and otherwise retain the log.VEventf? I think you can continue to use log.VEventf and then use level=0 for the analogous call to log.Infof, but please double-check if you decide to write this like that.

[nvanbenschoten]

This still seems like a good idea. I imagine we'll thank ourselves for adding in this logging at some point in the future.