[multi-stage] Query timeout overwrite

[61yao]

Currently, the query option time doesn't overwrite correctly.

We should have:

user-config override should always be considered as the total time (maybe minus planning time, i will explain later)

with this in mind, all the current timeout configuration during query should also be considered as total time all stages have a fixed timeout at an absolutely ts value = sys.currentTimeNano() + desiredTimeoutNano

if any stage hits this timeout, they will all agree to terminate.

See details in: https://github.com/apache/pinot/pull/9571#issuecomment-1278094979

[Jackie-Jiang]

Is this for multi-stage (v2) engine or also apply to v1? I believe v1 has the correct behavior.

[61yao]

This is for multi-stage only. Let me update the issue title