[HPA] support Pod custom metrics in autoscaler spec

[moh-osman3]

From internal customers, there have been multiple requests for a way to scale collectors based on Otel collector metrics. This would require exposing custom metrics to provide a way for users to configure their desired metrics beyond the resource metrics already supported. An example use case would be when custom metrics are not necessarily correlated with resource metrics. (e.g. ensuring latency of app is below certain threshold as a service requirement).