[EPIC] Release OML 3.0

[AlekseySh]

Here are the steps to achieve goal:

• Do all modality specific logic only in Dataset and Model. Rest of the classes have to be modality agnostic!
• Stop using keys to accumulate modality related information. Such information has to be processed inside Dataset.
• Introduce RetrievalResults class to store retrieval results. It has to work with any modality. Also: optimize memory by avoiding storing full distances matrix. Implement visualisation logic here to have possibility to draw something outside of EmbeddingMetrics (the modality specific logic of drawing a particular item has to be implemented in Dataset).

Checklist:

• [ ] all disabled tests are turned on
• [ ] code examples reworked
• [ ] colab is updated
• [ ] all configs in pipelines are updated
• [ ] benchmark metrics are reproducible
• [ ] check all sort and topk function in the code that rebuild predictions. They need to be moved to RetrievalResults.