fix: make ipamv2 metrics resilient to missing custom resource definitions

[rbtr]

Reason for Change:

The metrics observer added in #2970 fails fast if a source of metrics is unavailable. This prevents legacy metrics from being collected at all in scenarios like (as occurred) the ClusterSubnetState custom resource definition is not installed in the cluster.

In this change, the observer is made resilient to missing data sources or errors in the data collection, and will make a best-effort attempt to record the metrics even in case of partial data availability.

Issue Fixed:

Requirements:

• [ ] uses conventional commit messages
• [ ] includes documentation
• [ ] adds unit tests
• [ ] relevant PR labels added

Notes:

[rbtr]

/azp run Azure Container Networking PR

[azure-pipelines[bot]]

Azure Pipelines successfully started running 1 pipeline(s).

[nddq]

lgtm, although this behavior (potentially missing data sources/ errors that affects the metrics) is unique to ipam v2 right?

[rbtr]

yes, due to how ipamv2 pulls the CRs directly (since it doesn't already have them, like ipamv1 did)