Concurrent jobs per autoscale runner

Description

Limit doesn't behave the same on traditional and autoscale runners. On traditional runners it defines the capacity of that runner while on docker+machine runners it becomes the number of runners you want to run in total.

Reference:

Limits how many jobs can be handled concurrently by this specific token. 0 simply means don't limit. For autoscale it's the upper limit of machines created by this provider (in conjunction with concurrent and IdleCount).

https://gitlab.com/gitlab-org/gitlab-ci-multi-runner/blob/master/docs/configuration/autoscale.md#runners-options

This currently slows down builds on autoscaled runners as provisioning takes time.

Proposal

I would like to be able to specify the capacity of autoscale runners just like we can on traditional runners. We often don't need to provision new machines because existing machines have plenty of capacity left to run extra jobs.