Automatic merge from submit-queue (batch tested with PRs 50016, 49583, 49930, 46254, 50337)

[Federation] Make the hpa scale time window configurable

This PR is on top of open pr https://github.com/kubernetes/kubernetes/pull/45993.
Please review only the last commit in this PR.
This adds a config param to controller manager, the value of which gets passed to hpa adapter via sync controller.
This is needed to reduce the overall time limit of the hpa scaling window to much lesser (then the default 2 mins) to get e2e tests run faster. Please see the comment on the newly added parameter.

**Special notes for your reviewer**:
@kubernetes/sig-federation-pr-reviews 
@quinton-hoole 
@marun to please validate the mechanism used to pass a parameter from cmd line to adapter.

**Release note**:

``` 
federation-controller-manager gets a new flag --hpa-scale-forbidden-window.
This flag is used to configure the duration used by federation hpa controller to determine if it can move max and/or min replicas 
around (or not), of a cluster local hpa object, by comparing current time with the last scaled time of that cluster local hpa. 
Lower value will result in faster response to scalibility conditions achieved by cluster local hpas on local replicas, but too low 
a value can result in thrashing. Higher values will result in slower response to scalibility conditions on local replicas.
```