[GH-#] Error handling in RunWithTimeout

[Un1oR]

In this PR, two changes related to concurrent.RunWithTimeout:

1. Instead of panicking, we cancel the current context that the other functions are running with and return the original error from the entire function. This is needed to be able to occasionally terminate with an error, right now we only know how to crash.
2. We keep track of where the function that ended with the error was created. This is very useful for debugging, because unlike panic (inside a function) when an error occurs, we have no way to reliably trace the source of the problem. I think that overhead is not terrible here, because we only create long-lived functions once at the start of services, and do not do it all the time.