If for instance you have an instance which is failing to boot from as a result of some misconfiguration, or a botched upgrade, is it possible to mount a rescue ISO to fix it?
Sorta… If your instance is booted from a volume, then you can start a new instance and mount that volume to fix it. If your borked instance is booting from an ephemeral disk you will probably need to open a ticket to have one of our engineers try to rescue it.
A best practice is to consider the virtual machines disposable: keep their configuration in a config management system, like ansible or puppet, and the data in a volume, so if you botched an upgrade you just kill the machine and start a new one.
How do you boot directly from a volume? Is it available from the web Ui?
It is available from the web UI, from the drop-down menu pick the option Boot from image (create a new volume).
You may also want to read more about ephemeral and volume boot disks on the knowledge base: