feat: Adds the ability to select the runtime and inference service for playground

[bmahabirbu]

What does this PR do?

see above

thanks to @axel7083 for his work so far in #3077
I think ill be able to use InferenceRuntimeSelect.svelte to help select the runtime and pass it to the playground creation IF there is no model services available to select. If so i think ill just point the user to create the service in the model services page

Screenshot / video of UI

so far these are pics of selecting the inference provider

What issues does this PR fix or reference?

fixes #2613

How to test this PR?

create multiple services with the same model and see if you can choose from it in the playground

[jeffmaury]

Got to test it but this is not what I had in mind: for me when selecting a model, you should first select the runtime type (all / llama-cpp /openvino) and based on your selection the list of available models is filtered. It seems to me aligned with what is described in the issue but asking @slemeur for confirmation

[bmahabirbu]

Thanks @jeffmaury! Agreed I'll work on what you described in a separate pr.

Once that is done maybe we can revisit this idea later down the line once each model has multiple runtimes to pick from.

This maybe useful if a user has multiple services with different runtimes for a specific model they wanted to pick from.