Retry on failed creation due to not enough vCPUs

[bheight-prdca]

One issue that we've had with trying to use this action with the current MacStadium Blog instructions is that if you don't have enough vCPUs available to spin up a new runner it will hard fail and result in a PR failure due to no CPUs being available. Which means having to manually restart the failed action at a later time when more vCPUs become available to create a new orka ephemeral runner.

It would be nice if this action would execute with retries (rather than just failing) until enough vCPUs become available to create a new runner.

That sort of behavior would match the current hosted runner behavior of simply waiting until a runner becomes available for use.