chore: support customized OpenAI model.json

[louis-jan]

Describe Your Changes

This PR allows users to customize model.json to point to a compatible OpenAI endpoint. It applies the legacy model scanning for these cases.

Changes made

The git diff highlights the following changes in the model-json.ts file:

1. Added Local Engines Array:

   • A new constant array LocalEngines has been added. It contains specific inference engines from the InferenceEngine enum: cortex, cortex_llamacpp, cortex_tensorrtllm, cortex_onnx, nitro_tensorrt_llm, and nitro.

2. Modified Conditional Logic:

   • The conditional check within the scanModelsFolder function was altered.
   • Previously, the code returned the model if all existFiles evaluated to true.
   • Now, the code checks if the model's engine is not in the LocalEngines list or if all existFiles are true before returning the model. If either of these conditions is met, the model is returned.

These changes adjust how models are processed based on their engine type and the existence of corresponding files.

[github-actions]

Barecheck - Code coverage report

Total: 69.62%

Your code coverage diff: 0.00% ▴

Uncovered files and lines

File	Lines
web/hooks/useModels.ts	45, 58-59, 63