remove check for returning in-memory size when VMSS is in updating state

[marwanad]

We've had this logic for a while now. If the VMSS is updating state, return the in-memory size for a nodegroup/scale-set.

This was trying to guard from a scenario like below:

1. Autoscaler scales up to 4 instances
2. We get hit by a cache refresh
3. ARM hasn't registered our new capacity goal so we get stale data and we end up updating the in-memory cache to a lower value.

We've recently guarded against this case by this PR, where we extend our in-memory TTL so we should never hit that scenario.

For cases where the TTL is long, I believe this "Updating" logic can lead to a bad scenario where we take very long to reconcile with the real IaaS capacity goal if you for example get hit by multiple scale ups and it takes VMSS too long to scale or if cloudprovider is updating the network profile. In this case, it's best to not have it.

In the case where you're also experiencing rapid spot evictions, the VMSS state will be "updating" so you'll miss the chance of noticing a Spot eviction and reacting to it before the node object gets removed and pods evicted.

/area provider/azure

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: marwanad

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• cluster-autoscaler/cloudprovider/azure/OWNERS [marwanad]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[nilo19]

/lgtm