add MimirHPANeedsToBeScaledUp alert

[TheoBrigitte]

Towards: https://github.com/giantswarm/giantswarm/issues/31385

This PR adds MimirHPANeedsToBeScaledUp alert in order to detect when an HorizontalPodAutoscaler in the mimir namespace has reached its maximum capacity; meaning replicas are maxed out and resources usage is above targets.

For this alert I used the kube_horizontalpodautoscaler_* metrics provided by Kubernetes:

• first part of the query detects when the current desired replicas has reached the maximum allowed replicas
• second part of the query detects when the current metrics value used for autscaling (cpu or memory) is above target. When both conditions are met the alert fires.

[TheoBrigitte]

@QuantumEnigmaa should we change this to make it work for Loki also ?

[QuantumEnigmaa]

I think we can do that later. Let's merge this one and the loki one first so that we have those enabled asap and then let's create an issue to refactor this so that we can discuss how we want to do it

[QuantumEnigmaa]

Shouldn't you add UTs ?

[QuentinBisson]

LGTM apart from missing tests :)