Levenberg marquardt cleanup

[minnerbe]

I recently used the LevenbergMarquardtSolver in a project. Despite the experience being very smooth overall, I noticed some typos and inconsistencies in the documentation that made using the class a bit inconvenient. This PR tries to fix most of those; in particular, I:

• fixed typos and warnings indicated by my IDE;
• made two new methods computeSquaredError and fit(<7 arguments>) in LevenbergMarquardtSolver in which the order of the arguments is consistent with the order of arguments in the FunctionFitter interface and LevenbergMarquardtSolver's constructor;
• deprecated the methods with arguments in different order;
• precomputed the values and gradients of the function to fit to save some (potentially costly) function evaluations.

Let me know if you think that makes sense or if I can do anything else.

[tinevez]

Thanks! I am the original author of these files and approved the changes.
I don't want to break imglib2 policy. @tpietzsch can I merge these changes in master?

[minnerbe]

Great! I was really glad that this functionality exists within the ImgLib2 universe, so thank you for providing it in the first place 🙏