fix: re-raise RuntimeError when uvloop raises RuntimeError during connect

[bdraco]

What do these changes do?

Its possible that uvloop raises RuntimeError during connect. ie RuntimeError: File descriptor 59 is used by transport <TCPTransport closed=False reading=True 0x55a6ec5c9460>. Currently we don't trap RuntimeError as the code copied from asyncio.base_events assumed only OSError can happen. Its possible that selector_events can raise RuntimeError as well https://github.com/python/cpython/blob/e44eebfc1eccdaaebc219accbfc705c9a9de068d/Lib/asyncio/selector_events.py#L271 but it doesn't seem to happen in practice as the root cause appears to be a bug in uvloop. It may be that loop.sock_connect is not cancellation safe in uvloop and is leaking the socket fd allocation on cancel or something else...

The file descriptor reuse in uvloop is not fixed by this PR. This is something that must be fixed in uvloop itself

Are there changes in behavior for the user?

no

Related issue number

#93

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 100.00%. Comparing base (b075f25) to head (23f6bd4).
Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff            @@
##              main      #105   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files            6         6           
  Lines          198       201    +3     
  Branches        45        47    +2     
=========================================
+ Hits           198       201    +3     

Flag	Coverage Δ
	99.00% <100.00%> (+0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.