Update optim

[jobrachem]

Closes #194

The changes are:

1. The function now includes a progress bar
2. The function now avoids including the model state into the compiled functions

[jobrachem]

@GianmarcoCallegher yesterday, I found some more issues, so I converted the PR back to a draft and fixed them. It is now ready for review. The issues concerned:

1. The implementation of early stopping contained a thinko. I had compared the current loss to the best loss within the patience periord. Now I compared the oldest loss in the patience period to the best loss within the patience period.
2. The iteration counter and index handling was off by one. Now, iteration 0 refers to the initial values. Previously it referred to the values after the first update.

[GianmarcoCallegher]

@jobrachem I saw new commits. Is it ready for the review?

[jobrachem]

@GianmarcoCallegher Yes :)

[GianmarcoCallegher]

Well done 😄