feat: model provider configuration validation

[iwilltry42]

Important: This draft PR contains a couple of changes.

1. Change to the handler to populate ToolReference.Status.Error with an error message if available models couldn't be fetched
   
2. Change to the UI when calling /available-models/{model-provider-id} so it shows the error message to the user
   
3. Change to the UI to call /validate and only if that succeeds, save the config and call /available-models
   • If it fails, don't save the config and display an appropriate error
4. Add the /validate endpoint for model providers which calls the subtool validate from <model-provider-tool> to validate the config (if available)

Tested with the "reference implementation" which is the gemini-vertex provider: #986 / obot-platform/tools#297

Ref:

• Need consistent way to validate a model-provider is properly configured #842
• Model provider shows configured state (with no indication of error) after it is being misconfigured. #819
• Voyage - When voyage model provider is configured with invalid api key, there is no error message presented to the user. Also voyage models are made available to set default aliases. #993
• Anthropic provider - When Anthropic provider is configured with invalid api key , there is no error message presented to the user. Also anthropic models are made available to set default aliases. #991
• When model is configured with incorrect api keys we do not show the banner relating to model provider setup or restrict chatting. #974

[thedadams]

Does this work? It looks like thread.Name will always be empty here.

[iwilltry42]

Likely not - leftover from when I didn't use GenerateName