Create an endpoint that generates user mention suggestions

[acelaya]

We are currently building the mention suggestions list by combining users who already annotated the document, and members of the active group.

Then we sort and filter that list client side.

This has some limitations though, like not being able to suggest users who are not part of none of those two lists mentioned above. You can still mention them by manually typing their username, but that could be improved.

Instead, we might have a dedicated endpoint that, given a search term (the partial mention) would try to build a contextual list of users we think are more relevant for you or current context, including users you frequently interact with, users you have actively decided to follow, students which are part of your course (in the LMS), etc.

The list of rules to decide who ti suggest are not decided yet.

[robertknight]

Relevant Slack thread which discusses issues around duplicate h user accounts in the LMS context, arising from issues such as teachers impersonating students (using built-in LMS features), and how using the LMS roster data could avoid this.