Increase / allow configuration of openstack VM boot timeout

Hello,

A user on PS7 is experiencing test failures related to Openstack VM boot timeouts. I looked at spread and it appears the timeout is a static 2 minutes: https://github.com/canonical/spread/blob/ae284792596e00d325a1787604fe4ec7e00574aa/spread/openstack.go#L389

I think this is an unreasonably low timeout for spawning VMs on an Openstack cloud. We have had instances where VMs take > 5 minutes to enter an active/running state on the Prodstack clouds.

I think the following should be done:
- increasing timeout significantly (8 minutes)
- parameterizing this timeout to allow user configuration

Also, additional logic to check for BUILD versus ERROR state could be added to make the checks more robust. IIUC all VMs that fail to build should eventually transition to error state

([mattermost reference for the user issue here](https://chat.canonical.com/canonical/pl/xh7s883od7n7zba57wj4wcna7e))

Thanks,
Alex Lukens

Start in spread/openstack.go around line 389, where the static two-minute VM boot timeout is used. Trace how OpenStack VM settings are configured and how the current state is polled. Done means the timeout has a significantly higher default, can be configured by users, and handles BUILD versus ERROR states as appropriate.

Written by the indexing model from the issue text.