Add LiteLLM provider for OpenAI API compatible models

[mochow13]

The MR adds support for LiteLLM in Pydantic AI. The idea is to use a LiteLLMProvider for OpenAIModel since LiteLLM API is OpenAI API compatible.

Closes #1496

[DouweM]

What do you think about calling this provider_name like in HuggingFaceProvider?

[mochow13]

custom_llm_provider seems to be more aligning with LiteLLM's support for setting one's own provider.

[mochow13]

@DouweM Pipeline is failing due to:

Name                               Stmts   Miss Branch BrPart   Cover   Missing
-------------------------------------------------------------------------------
tests/models/test_model_names.py      68      6      8      0  92.11%   78-98, 140-141
-------------------------------------------------------------------------------
TOTAL                              28027      6   4472      0  99.98%

236 files skipped due to complete coverage.
Coverage failure: total of 99.98 is less than fail-under=100.00
make: *** [test] Error 2

I am not sure what to do. In the test_model_names.py file, there is no code related to OpenRouter or Together. So confused whether we should put any tests for LiteLLM or not.