Update some sensible defaults and documentation

[simonrw]

Motivation

When running LocalStack on EKS on Fargate, the LocalStack pod often gets terminated. This is for two reasons:

• the health probes are too restrictive, and
• the default pod requests cause pods to be terminated too quickly

Changes

A consideration was made for both local and EKS/Fargate scenarios, so defaults were not updated to handle one scenario at the cost of the other.

• Increase default probe delay and failure thresholds
  • these do make local development a little slower, but in my experience, LocalStack does not take much less than 10 seconds to become ready anyway
• Add section to the docs on helm customzation
  • the defaults were not updated because this may cause resource exhaustion for local development which would not be a good user experience

[alexrashed]

Awesome! Thanks a lot for jumping on this! The new defaults are less strict than the ones before, so this is a backwards compatible change.
Also thanks a lot for proactively updating the README with additional information on how to set the resource limits! 🚀 📓