better chat reporting in merge_pull_request

Python's requests library has logic for "if not response" that returns "false" for all 4xx and 5xx HTTP codes, including 404.

The CI testing calls test_approve() which calls assert chatcommands.approve(8888, original_msg=msg) == "Cannot connect to GitHub API" (for a non-existent PR), so the test/wrapper needs to be updated at the same time. What does: m.setattr("requests.get", lambda *args, **kwargs: None) do?

• looks like it's a python "do-nothing" command/function

look for / watch all the places that call approve(); seems to just be in gitmanager.py and test_chatcommands:

• assert chatcommands.approve(8888, original_msg=msg).startswith("You need blacklist manager privileges")
• assert chatcommands.approve(8888, original_msg=msg) == "Cannot connect to GitHub API"
• assert chatcommands.approve(2518, original_msg=msg)[:8] in {"PR #2518", "Cannot c"}

Test manually, somehow, with pytest and:

    with monkeypatch.context() as m:
        # Oh no GitHub is down
        original_get = requests.get
        m.setattr("requests.get", lambda *args, **kwargs: None)
        assert chatcommands.approve(8888, original_msg=msg) == "Cannot connect to GitHub API"
        m.setattr("requests.get", original_get)