File storage locations

[gschlafly]

We should clarify where all the relevant information and files are stored, or should be stored, when processing information. This includes the examples, presets, comments, metadata, calibration, and neural net model files. We can put this in the README.md, or in this issue thread in the meantime.

[amitabhverma]

The example datasets distributed with the plugin reside under the napari plugin installation which on Windows is :
..\anaconda3\envs\napari-env\Lib\site-packages\napari_lf\examples\
Each included dataset will contain a metadata.txt file with it's default parameters which get loaded based on the data directory on the napari-LF plugin.

A user would change this data directory and point it to their own data folder. The parameter values that exist on the plugin will remain since this new data location would not have a corresponding 'metadata.txt' file to load the data specific parameters. It will be the responsibility of the user to ensure the parameters are valid before proceeding with processing. @gschlafly - (should we have a dialog prompt to indicate this ?) Once a 'calibrate' (or rectify, deconvolve) is performed a 'metadata.txt' file is generated and stored within that data folder. All files created (calibration files, output image files) are stored in the same data location.

The Neural Net model files (*.ckpt) are stored based on the location set in the jupyter notebook training script which resides in the installation directory ..\napari_lf\lfa\main_train_neural_net.ipynb. Once a model has been trained, for it to be used it has to be copied to the data folder. @gschlafly - (thinking more about this, I don't think this is a good implementation because a user might have many folders with the same type of data but then has to copy the trained model file to each one of them).

Finally, the state of the plugin is saved to settings.ini that resides under the plugin installation directory. @gschlafly - (probably a Reset Settings button would be useful). Any settings and parameters changed on the GUI are immediately saved. This also applies to the presets that can be saved and applied to datasets which do not contain a corresponding 'metadata.txt' file.