OMP is less memory intensive + notebook improvements

[paulshuker]

Note:

This is not backwards compatible with previous run-throughs of the Omega branch. You must delete the notebook and re-run. However, you can keep extract/filter images in the tile directory.

Changes:

• The notebook now returns lists instead of tuples (#311).
• OMP now saves results in a zarr group within the notebook. Each tile is saved in a subgroup. This is less memory intensive and allows a large notebook with OMP to be loaded much faster. Also, the results are now compressed much better (#312).
• OMP default config is slightly better.
• The notebook has a more sophisticated config checker. Warnings are now issued about a modified config file only if that config section has already been used within the pipeline. This catches mistakes when the user thinks they have changed the pipeline by modifying the config file, but the config section has already been used.

Checklist

• [x] I have updated the changelog.txt, if applicable.
• [x] I have bumped the software version in _version.py, if applicable.
• [x] I have checked that the unit tests passed.
• [x] I have checked that Minnie and/or RoboMinnie passed.
• [x] I have resolved merge conflicts.

[reillytilbury]

These changes look great. I like the use of zarr groups!