Automate large scale load test

[jaypipes]

Similar to #691, this is for automating of a larger scale test with 100 nodes and 4300 pods:

1. create 100 c4.xlarge node cluster
2. scale replica from 0 to 4300
3. verify kubectl get deployment all to available
4. scale replica from 4300 to 0
5. verify kubectl get deployment

This should not be done on each PR of course, but set to run automatically before a release commit/tag and periodically

[mogren]

@bnapolitan For reference :)

[gyuho]

c.f. aws/aws-k8s-tester#91

This can be either added to aws-k8s-tester as a nested add-on or implemented inside amazon-vpc-cni-k8s. If added to aws-k8s-tester, we may not get the full control of when to scale up and down. If added to amazon-vpc-cni-k8s, code base will be messy with a bunch of vendoring.

[mogren]

Added in #1068