enhance background job management of region cache

[zyguan]

There are many background cron jobs in region cache. This PR tries to reduce boilerplate code (like the following) by bgRunner.

ticker := time.NewTicker(interval)
r.wg.Add(1)
go func() {
	defer func() {
		r.wg.Done()
		ticker.Stop()
	}()
	for {
		select {
		case <-r.ctx.Done():
			return
		case <-ticker.C:
			// ...
		}
	}
}()

ref: #1104

[cfzjywxk]

@crazycs520 PTAL

[cfzjywxk]

LGTM

[crazycs520]

How about moving this line after line 575? since the ticker is stop in another goroutine.

[zyguan]

My point is: if the interval <= 0, it panic when calling schedule, which is able to be recovered by the caller; however if we move it into the goroutine, panic cannot be caught by users.

[crazycs520]

If so, we should check interval <= 0 here, instead of causing panic and recover to handle it?

[zyguan]

IMO, panic on interval <= 0 is reasonable, but the panic should be recoverable. I will update it as you said anyway.

[crazycs520]

I think we should try to check in advance to avoid panic, Instead of letting it panic, after all, this is a system software, users will panic when they see panic logs.

[zyguan]

I've changed it.
BTW, panic let you kown you've passed a wrong arg, however, keeping silent sometimes hide potential risks.

[crazycs520]

ditto.