Generalize the types of errors caught

[AlanCoding]

I'm somewhat-frequently seeing this as a result of github flake.

E               File "/home/ec2-user/venvs/test-controller/lib64/python3.9/site-packages/pluggy/_callers.py", line 39, in _multicall
E                 res = hook_impl.function(*args)
E               File "/home/ec2-user/venvs/test-controller/lib64/python3.9/site-packages/pytest_github/plugin.py", line 333, in pytest_itemcollected
E                 self._issue_cache[url] = IssueWrapper(self.api.issue(username, repository, number), self)
E               File "/home/ec2-user/venvs/test-controller/lib64/python3.9/site-packages/github3/github.py", line 1176, in issue
E                 json = self._json(self._get(url), 200)
E               File "/home/ec2-user/venvs/test-controller/lib64/python3.9/site-packages/github3/models.py", line 216, in _get
E                 return self._request("get", url, **kwargs)
E               File "/home/ec2-user/venvs/test-controller/lib64/python3.9/site-packages/github3/models.py", line 206, in _request
E                 raise exceptions.ConnectionError(exc)
E             github3.exceptions.ConnectionError: <class 'requests.exceptions.ReadTimeout'>: A connection-level exception occurred: HTTPSConnectionPool(host='api.github.com', port=443): Read timed out. (read timeout=10)
E           assert False

If you consult the library used for github3, you can see that the ConnectionError is not a subclass of github3.exceptions.GitHubError which is what is currently caught.

This results in the test run being aborted with this error and something about an "internal error" in pytest. It also causes the pytest-mp to collect an inconsistent number of tests, which I can't fully understand now.

>>> import github3
>>> issubclass(github3.exceptions.ConnectionError, github3.exceptions.GitHubException)
True
>>> issubclass(github3.exceptions.GitHubError, github3.exceptions.GitHubException)
True

This change should catch both scenarios.