Instance create retry fails

This seems to be a combination of bugs that will happen in an up-to-date environment, the following occurs:

1. Heat creates a volume
2. Heat creates an instance and tells Nova to boot from that volume
3. Instance creation fails, however Nova already reserved the volume (so the volume is in attaching state)
4. Heat deletes the failed instance
5. Heat tries to recreate the instance, but the volume is in 'attaching' state (because Nova reserved the volume), so it throws this exception:

BadRequest: Invalid volume: Invalid input received: Invalid volume: Volume status must be available to reserve, but the status is attaching. (HTTP 400)

6. Due to the checks here, Heat does not use the first failure as the stack failure, but the second one because it is not an instance of `exception.ResourceInError`.
 This is because of the following: https://github.com/openstack/heat/blob/stable/pike/heat/engine/resource.py#L1178-L1179

I think this has to be mitigated in Nova but also Heat should probably report the first failure if there are subsequent failures in the retry mechanism, even if "BadRequest" was the failure.