1. Proposed title of this feature request

Configure Kubelet Image Garbage Collector Settings

2. What is the nature and description of the request?

Allow the settings imageGCHighThresholdPercent and 
imageGCHighThresholdPercent to be managed the same way that pod pid limits are managed by updating the kube config.
 
Examples:

• rosa create kubeletconfig -c example --image-gc-high-threshold-percent=75
• rosa create kubeletconfig -c example --image-gc-high-threshold-percent=60

3. Why does the customer need this? (List the business requirements here)

Customers may experience lots of DiskPressure issues on their nodes and need tunables to correct it.  This could manifest itself into a myriad of other problems (containers evicted, node issues, etc).  Upon investigation with a customer, we found that their container images are sometimes as large as 2GB and they are doing hundreds of builds/deploys per day as part of their CI processes, which is causing this issue.  The default Image/Container GC cannot keep up with this.

I did already have a look at the customer Dockerfiles for their apps and they are already doing lots of good things like multi-stage builds, slim images, etc, so I am not sure how much we can slim down their images (I am working on it).  I think customers may benefit from being able to tune a couple of these settings.  Specifically the imageGCLowThresholdPercent and imageGCHighThresholdPercent which would allow them to GC the images before hard evicting their containers at the imagefs.available percent (great article on that at https://www.redhat.com/en/blog/image-garbage-collection-in-openshift) as well as maintaining a lower disk space full of images to prevent hard eviction and DiskPressure issues.

4. List any affected packages or components.

ROSA, OCM, OpenShift, KubeletConfig