Fix CI concurrency

[Yury-Fridlyand]

Few CI improvements witch will reduce occupation CI queue and eliminate stale runs.

1. Kill CI jobs on PRs once PR branch gets a new push. This will prevent situation happened today - a huge job triggered twice in less than an hour and occupied all org (for all repositories) runners queue for the rest of the day (see pic). This completely blocked valkey-glide team.
2. Distribute nightly croned jobs on time to prevent them running together. Keep in mind, cron's TZ is UTC, so midnight tasks incur developers located in other timezones.

This must be backported to all release branches (valkey-x.y and x.y)

[enjoy-binbin]

thanks, the ideas are great

[madolson]

This is cool! I didn't know about this. It's a little unfortunate that sometimes it's nice to submit a couple of sequential builds, but we can also tune it if we need to.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 70.23%. Comparing base (aaa7834) to head (aae5920).
Report is 7 commits behind head on unstable.

Additional details and impacted files

@@             Coverage Diff              @@
##           unstable     #849      +/-   ##
============================================
- Coverage     70.35%   70.23%   -0.13%     
============================================
  Files           112      112              
  Lines         61467    61467              
============================================
- Hits          43248    43170      -78     
- Misses        18219    18297      +78     

see 13 files with indirect coverage changes