Automatic merge from submit-queue (batch tested with PRs 36693, 40154, 40170, 39033)

Refactored kubemark into cloud-provider independent code and GCE specific code

Ref issue #38967 

The following are the major changes as part of this refactoring:
- Moved cluster-kubemark/config-default.sh -> cluster-kubemark/gce/config-default.sh (as the config is gce-specific)
- Changed kubernetes/cluster/kubemark/util.sh to source the right scripts based on the cloud-provider
- Added test/kubemark/skeleton/util.sh which defines a well-commented interface that any cloud-provider should implement to run kubemark. (We have this interface defined only for gce currently)
  This includes functions like creating the master machine instance along with its resources, executing a given command on the master (like ssh), scp, deleting the master instance and its resources.
  All these functions have to be overrided by each cloud provider inside the file /test/kubemark/$CLOUD_PROVIDER/util.sh
- Added the file test/kubemark/cloud-provider-config.sh which sets the variable CLOUD_PROVIDER that is later picked up by various scripts (start-kubemark.sh, stop-kubemark.sh, run-e2e-tests.sh)
- Removed test/kubemark/common.sh and moved whatever provider-independent code it had into start-kubemark.sh (the only place where the scipt is called) and moved the little gce-specific code
  into test/kubemark/gce/util.sh.
- Finally, removed useless code and restructured start-kubemark.sh and stop-kubemark.sh scripts.

@kubernetes/sig-scalability-misc @wojtek-t @gmarek