If assigning 'inplace', and there's a value, assign globally

[brucemiller]

Patch following discussion in #2527. If we assign "inplace", with an actual value (as opposed to clearing), and there is no previous assignment, we'll assign globally. A reasonable default, if we end up ever needing the feature.

[dginev]

This is for @xworld21 to review, I would still prefer that code path disabled with a warning.

[brucemiller]

For some reason, I can't add @xworld21 to reviewr

[dginev]

@brucemiller same, I think the repository settings are too restrictive. In fact I know they are, because I don't have any access to them - you are the only one who can investigate there.

It's the "Settings" tab at the rightmost top of the repo page. See #2523 as a related problem to settings.

[brucemiller]

Ah, the docs are somewhat hard to parse, but I think the problem is that for free accounts, you can only request one reviewer. And I haven't discovered a way to UN-request the review from @dginev

[dginev]

That's not quite it, since you can't request external reviews on any of the other PRs that don't have a reviewer set either, for example #2364

[dginev]

PR code matches its description.

[xworld21]

Good that you didn't wait for my review: I don't understand frames at all...