workload/tpcc: fix N^2 row updates during long-duration reset

[ajstorm]

Summary

• Fixed a bug where the TPC-C reset worker goroutine was started once per worker instead of once total
• With 10 workers per warehouse by default, a workload with W warehouses would spawn (10*W) goroutines, each updating all W rows every 24 hours
• This change uses sync.Once to ensure only one reset worker is started, reducing the 24-hour reset from (10*W)*W to W row updates
• For a 1,200 warehouse cluster, this reduces the reset from 14,400,000 row updates to 1,200 row updates

Test plan

• Manual verification on a running cluster using curl http://localhost:33333/debug/pprof/goroutine?debug=2 | grep -c startResetValueWorker
  • Before fix: 2400 matches (1200 goroutines × 2 matches per stack)
  • After fix: should show 2 matches (1 goroutine × 2 matches per stack)

Informs: #124259

🤖 Generated with Claude Code

[blathers-crl[bot]]

Your pull request contains more than 1000 changes. It is strongly encouraged to split big PRs into smaller chunks.

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

[cockroach-teamcity]

This change is