config, doc, test: clarify nick@host format of owner, admins settings

[SnoopJ]

Description

This changeset clarifies in the documentation that the core config field owner may be given as nick@host as well as a nick, and also covers this previously-untested usage in the Trigger tests. (Edit: this applies to the admins field as well)

Checklist

• I have read CONTRIBUTING.md
• I can and do license this contribution under the EFLv2
• No issues are reported by make qa (runs make lint and make test)
  • 1451 passed, 8 xfailed, 1 warning in 47.36s
• I have tested the functionality of the things this change touches

Remarks for reviewer

In moving the dummy configs to a fixture, I originally wanted to move the template out of the module namespace and into the fixture, relying on textwrap.dedent() to take care of the indentation:

@pytest.fixture
def tmpconfig(configfactory) -> Config:
    return textwrap.dedent(f"""
        ...
    """

But this approach can cause trouble with trying to add a field later if also using an f-string, so I settled on a fixture that returns source. In particular, the thing I stumbled on was wanting to write:

def test_ircv3_account_tag_trigger(nick, tmpconfig_src, configfactory):
    line = '@account=bar :Nick_Is_Not_Foo!foo@example.com PRIVMSG #Sopel :Hello, world'
    pretrigger = PreTrigger(nick, line)

    # NOTE: only first line of tmpconfig_src has the expected indent level
    # so the result ends up exactly the same as without the dedent()
    tmp_config_account = dedent(f"""
        {tmpconfig_src}
        owner_account = bar
    """)

I don't feel strongly about eliminating the module-level TMP_CONFIG and if desired I can update this PR to preserve it.

[dgw]

In addition to the style tweaks mentioned as line notes, let's either include core.admins in this as well (it also uses match_host_or_nick()) or immediately follow up with another PR to do so.

[dgw]

Discretizing tests for the different cases will probably happen in whatever patch addresses #2734 (thanks for opening that). For now this is a good step forward.

10 commits is a lot, with the fixups and stuff, so this is your 👍 from me to squash things. 😸

[SnoopJ]

I have a smaller patch for #2734 that "just" adds a new test, will open that in a moment, but I don't think it's any big problem if we change the doc in one changeset and address the problem of test coverage in another, since that gap has been there a while anyway.

[dgw]

GH doesn't support relationships between PRs and issues (why?) but consider this blocked by #2736 for now. (@Exirel you're still welcome to review; certainly the tests are useful unless we get rid of the behavior entirely.)

[dgw]

After what one might call much discussion on IRC, we seem to have an alternative plan involving a new config value and deprecating certain value patterns of the existing owner setting. I will describe it in #2736, perhaps mirrored in #2587 (or just linked from there) as it's all related.