Design - TuningComponent

## Reasoning

It is not clear that AutoMLSearch is tuning the pipeline threshold during search.
- It's not mentioned in our docs: https://evalml.alteryx.com/en/stable/user_guide/automl.html
- It's not mentioned in the logs when `verbose=True`

Since it is not clear that threshold tuning is happening, users can be confused as to why the score on the best pipeline differs from the score they get when they call `fit` and `score` separately.
- See #2844

I think if threshold tuning were a component in the pipeline, it'd be very clear to users how and when threshold tuning happens during pipeline fitting. It also means users don't have to remember to tune the threshold themselves when they try to fit a pipeline "exported" from `AutoMLSearch`.

I'm interested in hearing other approaches but I think making it a component in the pipeline, similar to what we did for under/oversampling, is the best way to make it very clear that threshold tuning is taking place.

Closing Criteria:
- [ ] Binary classification pipelines have a threshold tuning component in the pipeline
- [ ] Docs describing threshold tuning