Add ML procs for Azure OpenAI service & AWS SageMaker Endpoints

[jexp]

Probably just a config setting for the regular openai service?

in python it needs:

openai.api_type = "azure" #I use Azure OpenAI. You might need to comment if you use the non-Azure instance
openai.api_base = os.getenv('OPENAI_API_BASE')
openai.api_version = os.getenv('OPENAI_API_VERSION')
openai.api_key = os.getenv('OPENAI_API_KEY')```

For Sagemaker, figure out the endpoints via langchain?


https://api.python.langchain.com/en/latest/modules/embeddings.html#langchain.embeddings.SagemakerEndpointEmbeddings

[jexp]

Azure openai should work the same as OpenAi endpoints, they just have an extra query string for API version.

Either add a specific implementation for it, where the user can also pass in the version as config.

Or reuse the openai implementation change it to support a placeholder for {path} in the URL base-string, instead of concatenating it just at the end of the URL:

so for an MSFT API like:

https://xxx-openai.openai.azure.com/openai/deployments/xxx-gpt-35/completions?api-version=2022-12-01

consider using a base-url like:

apoc.ml.openai.url=https://xxx-openai.openai.azure.com/openai/deployments/xxx-gpt-35/{path}?api-version=2022-12-01

for openai then use

https://platform.openai.com/docs/api-reference/{path}

but still support the old way if no {path} placeholder is in the base-URL