Z-drift and XY stabilization

[edyoshikun]

This PR has the code for:

• [x] Calculating the z-drift and returning a z-translation matrix for timelapses
• [x] Calculating the xy-translation and returning a xy-translation matrices for timelapses
• [x] CLI call for running the stabilization

The PR can be merged to registration_v2 or to main after registration_v2 is merged.

[edyoshikun]

@ieivanov @talonchandler this is ready for your input. I was using this small timelapse from the opencell, but feel free to pick a different timelapse. I did some cropping

mantis stabilize-timelapse -i /hpc/projects/comp.micro/mantis/2023_11_01_OpenCell_infection/1-recon/opencell_timelapse_7_phase.zarr/0/1/000001 -o ./test_stabilization/ -j 4 -c 0 -y -z -v -s 200 200

EDIT: mantis stabilize-timelapse -i /hpc/projects/comp.micro/mantis/2023_09_22_A549_0.52NA_illum/1-recon-phase/A549_MitoViewGreen_LysoTracker_W3_FOV5_1_phase.zarr/0/FOV0/0 -o ./test_stabilization/ -j 4 -c 0 -y -z -v -s 200 200

Couple of things to note that we can change and require your input:

• I don't currently have a config file because I wasn't sure if we truly needed one or if we could set the values through the CLI.
• Here, I am saving the arrays a npy. for both the z-drift and the xy-translations. Should we spit them out into a yml? The only advantage is that we can actually open a yaml file and somewhat inspect the matrices.
• The output is saved into a folder rather than a typical .zarr as we do in other pipelines. This folder will contain all the outputs (i.e npy files, png files, csv and the zarr). Do we want to change this?
• This function uses it's custom mantis.utils.apply_stabilization_over_time_ants() vs the one we use for deskew, label-free recons, registration process_single_position(). We can rename the functions.

[talonchandler]

Off-the-cuff thoughts based on out quick paired session:

• let's test on a dataset that actually needs stabilization
• I'm fine without a config for this case...no major preference
• I'll vote for human-readable transformations over .npy
• mantis.utils.apply_stabilization_over_time_ants() is fine for now IMO

[edyoshikun]

This is being replaced by #108