These nudges happen when a thread is within 24 hours of its auto-archive
time frame, and nudge the participants to take action to close the
thread out.

The current iteration gives the option to immediately archive, ask for a
decision, ask for a status, or ask for a task to be captured. The three
latter actions prompt for a person to identify as the person to decide,
give status, or capture task. The robot then pings that person and lets
them know that the thread will be auto-archived in 24 hours and to give
their info in that time frame.

Notably, by this point the bot has already set the thread to auto-
archive in 24 hours.

Soon we'll want a couple of these paths to re-extend the auto-archive
time frame back to the server default if possible, once the designated
person replies.

There is also a currently-unused code path that heuristically determines
if the full lifetime of a thread has been synchronous; in that case, it
does not act in any way to change auto-archive or prompt for outcomes.
These threads are treated as representing complete conversations.

When a user responds to the follow-up action request, they can designate
themselves or another user to take the follow-up action (e.g. capturing
a task, posting a status, or capturing a decision). This path looks
roughly the same for all actions, including pinging a user if the
requester selects someone other than themselves to take the follow-up
action.

Also add a FIXME regarding forced thread archiving.

Permitted features are tracked for a thread, channel, or category. The
functionality for whether a thread is sync or not is rolled into a
generic channel metadata functionality, which now includes permitted
features per thread/channel/category.

Note that there is not yet an external way to add to the permitted
features metadata.

This resolves an interaction error triggering if the user tagged themselves, so instead we wipe the dropdown afterwards.

to-do: Add channel for testing based on feedback