Explore using latest Karpenter

[T-Kukawka]

Context

Up till this point the ignition data was a dynamic set which was bringing troubles for the Karpenter implementation. We are changing the ignition to be placed in S3 bucket due to the length limitations: https://github.com/giantswarm/roadmap/issues/3442

With this change, there will be way less effort to support Karpenter and we can start experimenting with the latest version that brings a lot of useful features and improvements from upstream.

Acceptance criteria:

• [ ] With latest cluster-aws release, test the latest upstream Karpenter implementation
• [ ] Integrate the latest Karpenter into the default solution that GS provides

[paurosello]

Due to the current implementation of S3 bucket for MachinePools this will not work as we are creating new files every 15 minutes instead of updating a single file.