adopt orphan vms after a DB restore

[valeriap]

In order to improve the bosh administrator experience in managing disaster recovery scenario that requires a DB restore and to minimize the impact on customers, we propose to add to the BOSH Director and to the underlying CPI the possibility to adopt a list of vms.

[cppforlife]

cc @voelzmo

[voelzmo]

@cppforlife @valeriap I don't know if VMs should become important to bosh, so far they are just ephemeral things which you shouldn't care about.

With bosh deploy my-release followed by a command such as bosh attach disk job_z1/2 disk-1234 you could achieve the same. I'd actually prefer it this way. What do you think?

[valeriap]

@voelzmo
On the one hand I think that some vms can be important basing on their role in the deployed env; the importance relates to the impact on a customer when the vm is down.
On the other hand the proposed approach is easier and less error prone from an administrator point of view especially if you deployed a consistent number of vms that you have to reconciliate.

[valeriap]

cc @maximilien @aalbanes