Fix hot swhkd loop

[pastamoe]

This should resolve #317.

The actual fix:

• add delay to socket-connection attempts, so we don't cause unnecessary CPU load (this required pulling up the cooldown-calculation logic).

Misc changes:

• removed the "waiting for..." log, as it's kind of redundant with the log before the loop + the next change
• socket connection errors are no longer ignored, instead a warning with the cause is logged. I think 'warning' is fitting here, as this error results in a retry by the loops that call this function
• add more context to panics during set_uid/_pid calls. The original panic message is only useful with code at hand.
• get_uid now returns an error in case loginuid is unset, instead of running into calls like set_uid with invalid values (which then cause a panic)

The misc changes are of course not relevant for the referenced issue, but they helped me a lot to understand what's wrong with my systemd service setup. If you're unhappy with those changes, I can of course drop them.

Side-Note regarding the "socket connection warning": It's kinda bad that the log-level down there is "dictated" by how the loop further up behaves. I think it'd be better to pass an (enriched) error up, and let the loop decide what to with it. But I didn't want to introduce more changes, also because this is my very first time looking into Rust.