PyHealth processors -> embeddings not clear which are compatible

[jhnwu3]

We either need to update our documentation or make it more streamlined somehow.

• Documentation, and doc strings, and typing checks

We need to split EmbeddingModels into separate embedding models for easier readability.

1. same processor can have multiple potential embeddings (yes)
2. same embeddings can support multiple processors potentially as well.

The model itself should define processor compatibility and embedding types.

For concare, adacare, etc. we can simply use our existing adaptive approaches.