AWS instance size based on combination #69
Description
From experience, we found out that some combinations are slightly more resource hungry than others. When we deploy to AWS we default currently to c6a.xlarge, some combinations are behaving better on c6a.2xlarge, others (like Teku VC) require huge machines like c6a.12xlarge.
It'd be nice if we add some logic in kurtosis_aws_runner.py that will spin up different machine based on our observations. This will speed up our testing and make it less of a hassle.