Check goroutines in places related to recent found leaks

[jsoriano]

Follows the strategy used in #10850 to check for goroutine leaks
on tests in some places related to the leaks investigated as part
of #9302.

Several things here:

• New helper to reuse the goroutine checker, it also prints now the
  goroutines dump on failure.
• Add goroutine checks for autodiscover tests
• Add goroutine checks for CloseOnSignal and SubOutlet
  tests (they detect the issues solved by Stop waiting for signals on closed outleters #11263)
• Add goroutine checks for log input tests (they detect issues solved
  by Fix goroutine leak on initialization failures of log input #12125 and Fix goroutine leak on non-explicit finalization of log inputs #12164)

[andrewkroh]

Do our runtime metrics that get logged every 30s include a gauge for the number of goroutines? That would be helpful for identifying if a leak is introduced to a Beat.

[jsoriano]

I am moving the fix for the leak in NewInput to its own PR (#12125), so we can continue discussing about these tests here without blocking this fix.

[jsoriano]

Do our runtime metrics that get logged every 30s include a gauge for the number of goroutines? That would be helpful for identifying if a leak is introduced to a Beat.

Created PR for that #12135

[jsoriano]

Failing tests caused by #12171

[jsoriano]

@exekias @andrewkroh this would be ready for review, thanks!

[jsoriano]

Test failure in Windows looks related, I'll look at this

[andrewkroh]

Nice. That's helpful.