[kube-stack] Increase opentelemetry-operator resource memory limits

[rogercoll]

The operator pod currently defaults to 64Mi memory request and 128Mi memory limit. We've observed memory usage spikes reaching up to 200Mi, which has led to OOMKilled events.

To improve stability, should we consider increasing the memory limit to 500Mi, in line with the other deployed pods (e.g., collectors)?

[TylerHelmuth]

In the other charts we've started removing default resources and adding a warning to users to apply there our. We could follow that pattern here as well.

[rogercoll]

In the other charts we've started removing default resources and adding a warning to users to apply there our. We could follow that pattern here as well.

Sounds like a valid solution to me, there are many factors involved when deciding those defaults, and given the flexibility of the chart those values should differ for each configuration (collector binary/distribution, configuration, collector role, etc.).