invert dialogue time in ted

[Ghostvv]

Proposed changes:

• invert dialogue sequence before passing it to the transformer so that the last dialogue input becomes the first and therefore always have the same pos encoding
• fixes: Rasa 2.2.2 core model predictions off #7658
• fixes: Rasa 2.0.x fails to use stories where the previous events in the current conversation are different to the previous events in a story when making predictions #7221
• fixes: https://forum.rasa.com/t/ted-fails-overfitting/39474

Some results on Sara:
main -> conv acc 0.533, action acc 0.841
mcfly-ted -> conv acc 0.564, action acc 0.873

Status (please check what you already did):

• added some tests for the functionality
• updated the documentation
• updated the changelog (please check changelog for instructions)
• reformat files using black (please check Readme for instructions)

[dakshvar22]

Small refactor comments. Also, since you posted some results on slack, just for the sake of record can you post the results of the experiment in the PR description too?