Implement protections to minimize risk that a panic in a scaler impacts the operator execution

[swgriffith]

Proposal

I recently came across and documented the following bug in which the Github Runner Scaler can cause a panic with a simple typo in a parameter in the scaled job manifest. This panic put the KEDA operator itself into a crashloop. It would be good to add some protection in the call flow to capture panics introduced by a scaler and avoid impacting the operator as a whole.

#6156

Use-Case

Wrap the call to the scaler such that a panic within the scaler code is trapped, logged accordingly and the operator continues executing any other scaling activities on the cluster.

Is this a feature you are interested in implementing yourself?

No

Anything else?

No response