codl
exponential backoff should reset upon successful login
oftentimes a big backoff means an instance was offline for a while, if the user successfully logged in then it's probably back
of course there are cases where it's failing for some other reason, maybe some route is down, maybe we're being rate limited in nonstandard ways, idk, but having exponential backoff reset once is not a big deal
on second thought, rather than do it automatically, it might be best to tell the user "We've been having trouble reaching blabla, we'll keep trying later, or you can click here to retry immediately" (which would reset the backoff and retry not immediately but within the next few seconds)
Also might want to hide "next delete in" if we are backing off. Maybe only past a certain level of backoff.