provider: resume reprovide operations #1159
Description
Current situtation
A fresh reprovide cycle starts as soon as the SweepingProvider is created.
This means that if the node has been providing (and reproviding) content, then is turned off (not just offline) for a few hours/days, and then is restarted, it will take up to reprovideInterval for all keys to be reprovided.
Ideal situation
We want a similar behaviour to when the provide goes offline, which is to keep track of regions that should have been reprovided but weren't because the node was offline, and reprovide them as fast as possible when the provider comes back online.
We need to write reprovide timestamps to a datastore, so that when the SweepingProvider is started it can look in the datastore for regions that haven't been provided since at least reprovideInterval and schedule them to be reprovided ASAP.