LSpaceTest: make testEntryEventuallyRemovedAfterWaitersDrain more predictable

[barspi]

testEntryEventuallyRemovedAfterWaitersDrain was failing randomly because of thread timing chaos.
It failed more consistently in github actions. It failed less often on my fast machine.

This PR:

• introduces a new latch to make sure LSpace operations happen in a more orderly manner.
• a separate assert to test whether the lsp.in(key) got the correct value after asserting that it got some value.

Now it has passed the test on github. And after several hundred runs, it hasn't failed on my machine either.

[alcarraz]

Hi @barspi, have you considered awaitility for solving this kind of issues?

Just so we don't have to put arbitrary waiting times or code like the countdown latches. We would need to add a test dependency, but it would be useful for other tests as well.