Cloud Hypervisor: Improve analysis of failed starts

[ericooper]

Which feature do you think can be improved?

In the unlikely event the cloud-hypervisor doesn't start properly it is at present not possible to see at runtime in the log or journal why this happened because stdout and stderr aren't captured for the cloud-hypervisor fork/exec.

How can it be improved?

In golang it is supported to capture the output in a bytebuffer and use the content of the bytebuffer to print a suitable logging messsage supporting analysis of the root-cause as it should.

Additional Information

I've prepared a fix to be included in the next upcoming PR so this one can be assigned to me.