Bug: A VM having running status doesn't necessarily mean that the VM is actually running

[shayanh]

Environment

everywhere

Steps to reproduce

To make it more explicit add time.Sleep(10 * time. Second) to neonvm/runner/main.go somewhere before executing the QEMU command.

Actual result

NeonVM controller uses the runner pod's status to determine the VM's status. If the runner pod status is running and the VM's status is pending, then the controller changes the status of the VM to running. At this point, we don't necessarily know if the QEMU process has started, nor has the VM booted.

Expected result

Update the VM's status to running once is actually booted and running.

[Omrigan]

Copy from #811:

Implementation here depends on the definition of "VM start":

• The process start is one reference point
• Reaching a certain stage in the boot process is another
• Having a signal from userspace that some parts of the system started working is the third thing

We can ping the VM, either:

• our own service vm_monitor/comute_ctl
• [guest-ping through QGA], but here we have to bother with running QGA inside the VM