Add possibility to load pretrained embedding models

[egoriyaa]

🚀 Task description

Add possibility to load pretrained embedding models

Plan

1. For TS2VecEmbeddingModel and TSTCCEmbeddingModel classes modify load method in such way:

   def load(cls, path: Optional[pathlib.Path] = None, model_name: Optional[str] = None)

   • both parameters are specified - download the pretrained model and save it to path
   • path is specified, model_name is not - download the local model from path
   • path is not specified, model_name is specified - download the pretrained model to the etna folder in the home directory
   • both parameters are not specified - raise the exception

2. For TS2VecEmbeddingModel and TSTCCEmbeddingModel classes add static method list_models

   • Return list of available models

3. For TS2VecEmbeddingModel and TSTCCEmbeddingModel add parameter freezed in constructor. - If freezed is True, freeze model in constructor.

4. Modify 210-embedding_models notebook to show how to use pretrained models

Test cases

1. For load method:

   • 4 combinations of parameters
   • model_name not in list_models

2. list_models returns right model names

Additional context

No response