Error handling of missing environment variables

The actual handling of the exception is up to the different applications to handle gracefully - we should look into this more for the modules that comes with the AI module. The only one that is known to be graceful, is the API Explorer because we have put effort into showcasing errors correctly there on the actual chat message.

With that being said, there are two obvious things we can do to make this process easier globally:
* The list model command could come with a command to gracefully fallback if the api can't be reached and answer empty.
* A better exception in this case that actually tells you what is wrong.

This part can now be tested in the 1.0.x-dev:
* The list model command could come with a command to gracefully fallback if the api can't be reached and answer empty.

It shouldn't create wsod when the model response doesn't work, instead it will just give back an empy models selection.

@mErilainen - now both the things should have been added to 1.0.x. It fails gracefully when its loading the models, and the API is not responding or the key is set wrong on the AI API Explorer, with a log message.

And if the key is missing, similar to your case it now throws an AiSetupFailureException with the message "We could not load your OpenAI Key at all, please check your environment settings or your setup key.".