Update annotation and affixation handling

[clemera]

See #286, allow for more complex affixation items and refactor annotation code.

[minad]

Could you document your findings somewhere please? This would help for marginalia!

[clemera]

Okay, I will explain things in the docstring of the function.

[minad]

@clemera I pushed preliminary support for affixation function to marginalia. The affixation function is just a simple wrapper around the annotation function for now, see minad/marginalia@ef4fabf. I wonder about the docstring of selectrum--affixate, it sounds as if you can manipulate the candidate itself? This is not the case right?

[clemera]

Nice!

I wonder about the docstring of selectrum--affixate, it sounds as if you can manipulate the candidate itself? This is not the case right?

I think for display purposes affixation does allow that.

[minad]

And then it adds a 'display over the candidate or how is this supposed to work? How did you implement it in selectrum?

[clemera]

I think they take the strings as returned by the affixation use them for insertion into the completion buffer. In selectrum I use them for display as they are returned, too. For Selectrum the candidate index is used to retrieve the actual candidate so it doesn't matter how it is displayed, not sure how they handle it for default completion.

[minad]

Makes sense, thanks!