[BUG] parallel salt-cloud libvirt creation fails

Description

Creating multiple instances in parallel with salt-cloud using libvirt fails.

Steps to Reproduce the behavior

salt-cloud -m /etc/salt/cloud.map.conf -P vm1 vm2

[DEBUG   ] libvirt error (1, 18, "internal error: pool 'system' has asynchronous jobs running.", 2, 'internal error: %s', "pool 'system' has asynchronous jobs running.", None, -1, -1)
[INFO    ] Cleaning up after exception
[DEBUG   ] volume xxx.qcow2 cleaned up
[DEBUG   ] libvirt error (1, 18, "internal error: pool 'system' has asynchronous jobs running.", 2, 'internal error: %s', "pool 'system' has asynchronous jobs running.", None, -1, -1)
[INFO    ] Cleaning up after exception
[DEBUG   ] volume yyy.qcow2 cleaned up

Expected behavior

If VMs can not be created in parallel with libvirt and salt-cloud, this should be:

• documented; and
• automatically prevented, either by ignoring -P for this provider, or exiting with an error if -P is provided.

If VMs can be created in parallel with some improved code, then ideally it would be implemented.

Version

Second-hand report using 3002.

Additional context

In fairness, the salt-cloud docs do warn that parallel creation may not work for all providers.

Start by reproducing the failure with the salt-cloud -m /etc/salt/cloud.map.conf -P vm1 vm2 command and the libvirt provider, using the reported asynchronous pool error as the symptom to trace. Review the salt-cloud parallel-creation path and provider documentation; the work is done when parallel behavior is either supported reliably or explicitly prevented and documented.