I've seen various information about how, when using the Tor browser, it uses a different circuit for each domain, and (possibly) automatically generates a new circuit for the same domain every N minutes. However, I can't seem to find documentation for this (links would be appreciated!).
Some questions about the Tor CLI client though (not the browser):
Are these same new-circuit features from the Tor browser also enabled? Or, does it only ever generate a new circuit if instructed to do so through the control port?
If a node in the current circuit becomes unavailable (leaves the network), will it automatically detect this and generate a new circuit?
If it generates new circuits on its own (for either 1 or 2 above), does it publish anything on the control port, or is there any other way for an external application to be notified that the circuit has changed?
Again, any documentation for these questions would be super helpful.