Stop table-manager loop early, if it is done

[pstibrany]

This is trivial fix that helps integration tests to run slightly faster, as discussed at #2016 (review).

If table-manager is stopped during initial wait, it now exits early.

[pracucci]

Thanks @pstibrany. This PR looks based on a fairly old version of master. May you rebase it, please?

[pstibrany]

Thanks @pstibrany. This PR looks based on a fairly old version of master. May you rebase it, please?

Done. How did you even find out? You had to manually check my source branch I guess :)

[pracucci]

How did you even find out?

I checked out the PR to test it. I can see a failed integration test in CI. May you take a look, please?

[pracucci]

Thanks!

[pstibrany]

Thanks!

Worth noting: I haven't fixed the failing integration test. I think single-binary test with memberlist is flaky. I just added observed value into the error message. That said, I think it's unrelated to this change.

[pracucci]

Worth noting: I haven't fixed the failing integration test. I think single-binary test with memberlist is flaky. I just added observed value into the error message. That said, I think it's unrelated to this change.

Sure. But the log you've added may help to investigate it, as soon as we're able to reproduce it.