Serving runtime not properly configured if deployed after TrustyAI service

[ruivieira]

When serving runtimes are deployed after the TrustyAI service CR is applied, in some cases it might not be properly configured.

[alexcreasy]

I'll just add some information as to how I'm reproducing this from the dashboard perspective:

1. Create a new Data Science Project (i.e. k8s namespace)
2. Go to the settings tab and click Enable TrustyAI
3. Create a Data Connection - I'm using the rhods-public S3 bucket
4. Create a Model Server
5. Create a Model (i.e. an InferenceService in k8s)
6. Curl data at the model.

When it's done in the above order the curl commands error out. If you perform this sequence but move step 2 to last, everything works as expected.

[ruivieira]

This scenario is fixed with #118