Self Hosted Runners Optimisation

[AM1729]

Hi! So, we are running self hosted Runners, and sometimes there are some Runners which remain idle but we don't want to keep it that way, since it leads to resource wastage.

So, we thought of solving this issue by checking _work directory has been created or not after some time (say 5 or 10 mins) since we don't expect jobs to take more than 5 mins to be placed on a particular Runner.

However there is this issue:
Let's say after 5 mins just about when we would killing the Runner Process and Deregistering the Runner from the Org, now it says that The job is about to start on Runner: <About to be killed Runner Name>

So, in this state when the job is about to start and has not started yet , the work directory is not created, and hence internally it thinks that no job is running and ends the process.

Any suggestions about how to go about this issue ?
One thing that comes to mind is making an API call and check if the Runner (which we are about to kill) is in Idle or Active state.. however I wanted to know if when the job is having this state: the job is about to start does Github show the Runner to be in Idle or Active state ?

Please share your thoughts on this ...