Feature/dataset entry

[CloseChoice]

closes #2708

Add pydantic basemodel class (equivalent to dataclass but with stronger guarantees) to return from dolly dataset. Add the formatting functionality in the dataset entry class. This PR does quite a bit:

• add pydantic dependency
• introduce a new DatasetEntry class, which provides a method to do the formatting based on the mode and the QA_SPECIAL_TOKENS and the eos_token. This class should work as a general pattern to store and format single dataset entries. This class should also remove all the formatting errors we had previously with the datasets.
• add tests for DialogueDataCollator. I trained a minimal tokenizer only on the tokens that are present in the tests to not bloat the code (still a lot of LOC).
• added tests for the interplay of DialogueDataCollator and the newly introduced DatasetEntry class.
• add small fixes to be backwards compatible in handling the new DatasetEntry and old rows from the dataset.

todos:

• [x] mask system
• [x] remove changes in config
• [x] write tests for formatting of dataset entry