Make movie reviews model easier to run

[loostrum]

Currently it needs quite a lof of code to run; most of that can be made part of the model itself.
This would allow for easy use of the model in tests as well

[loostrum]

Leaving as-is for now, we should talk to an NLP expert first to see how models are normally used.

[loostrum]

The input to the model should remain text, but the functions to create tokens from the input string will be made part of the model. The goal is to avoid having to define a lot of boilerplate code in the dianna repository just to be able to run the model.

[loostrum]

Draft PR in the exploration repo: dianna-ai/dianna-exploration#147
The pytorch model is working, but after conversion to onnx it fails because there are no inputs.
Possibly related issue: pytorch/pytorch#44299

[loostrum]

We will not use the simplified model as it doesn't work. This issue can be closed when the simplifications that don't affect the onnx model itself are merged in dianna-exploration.

[elboyran]

@loostrum could you, please, check if this can be closed?