Use Case: Computer Vision e2e

[aguschin]

Right now, MLEM doesn't support working with images properly (only as arrays/tensors if we talk about mlem.api.save(..., sample_data=...)). Supplying just images there won't work as expected. This is a common scenario that we should support.

End2End CV Use Case would include:

1. Saving a CV model with mlem.api.save()
2. Deploying it with $ mlem deploy
3. Getting predictions from the deployed application with $ mlem apply import: Support image as input type. #310

Smaller tasks that contributes to this:

• Binary Serializers #447
• Improving serializers #509
• Add docs about this
• Write a blog post