add llama index's ollama embedding and fix type hint of embedding_model

[rjwharry]

Resolve #1070

• Add llama index's OllamaEmbedding class.
• Separate original add_from_dict function to add_from_dict and add_from_list. Because of get_param_combinations, modules will get embedding_model config as dictionary not list. But vectordb will get it as list. So I separate to two functions to handle both cases.
• Unify the way of loading embedding model.

[vkehfdl1]

LGTM Thanks:)