Strain mapping improvements: adding independent optimizers for model fitting and multipeak fitting for cepstral calculation

[mitis1]

What problem this PR addressees

This PR updates the fitting-models to allow for strain mapping and independent optimizers for each component. This allows each component to have a different learning rate, which improves model fitting.

It also updates strain_autocorrelation to support multipeak cepstral transform fitting for improved strain mapping. This is controlled by the refine_all_peaks option in choose_lattice_vector() and fit_lattice_vectors().

It adds a StrainFitting class in strain_visualization. This takes the u and v arrays from different strain calculation algorithms and uses them to calculate and plot the strain. The goal is to make the final plots look consistent regardless of which algorithm was used, and the results can be saved using Autoserialize.

What should the reviewer(s) do

See the test notebooks and verify they work. Also see the StrainFitting class and if it is usable for multiple different strain calculation methods.

Test notebook for multi-peak cepstral fitting and strain plotting using StrainFitting:
20260527-pmos-cepstral-strain.ipynb

Test notebook for model fitting and strain plotting using StrainFitting:
20260512-pmos.ipynb

[bobleesj]

@mitis1

I'm thinking electronmicroscopy:fitting_models_clean probalby need to be synced with dev so that this PR doesn'have to have 30k lines of code.

[bobleesj]

We want each PR should be meaningful and useful both for future us and this PR reviewers, I hope.

[bobleesj]

@mitis1 Please merge commits from upstream/dev into electronmicroscopy:fitting_models_clean

[bobleesj]

I saw #195 - if unsure, pls DM, etc. The goal is having this PR containing ideally mostly miti's commits.

[mitis1]

I just merged dev into fitting_models_clean, but I have some commits from nanobeam which I am working on removing so it is just additions to fitting_models_clean.

[bobleesj]

Seems like MAPED/polar stuff got in the way in this branch - meaning it has nanobeam stuff in this branch. This isn't a hygine PR but better tha 30ks and mostly relevant to fitting, so seems OK

[bobleesj]

But my second thought is that your branch contains upstream/nanobeam (like MAPED/PDF code being updated)

This is going to be an issue predicted if your changes need to be merged to upstream/dev eventually.

A clean case would be

1. Close this PR - make a backup of your files
2. Miti makes a new branch from electronmicroscopy:fitting_models_clean
3. Miti call it mitis1:strain-fitting-multi-optimizer
4. Commit a few ONLY related to your strain/optimization changes.
5. Make a new PR