WIP: memory checks, chunking, and new setting

[TimMonko]

References and relevant issues

When trying to load an 18MB (on disk) int32 label image I noticed a few things:

1. it's in-memory size is >8GB (data.nbytes)
2. it was being read as an image, despite its name containing labels, which I want for it to be read as a Label
3. it both loads in REALLY slowly (as in memory, because its uncompressing) and if it was lazy loaded...
4. slice-by-slice navigation of a TIFF was excruciatingly slow. I realizing this was because it was chunking by ZYX so each call to the slice was having to uncompress the entire file each time.

Description

1. Detect labels from file names, if channel names aren't enough
2. Change memory check to be the would-be in memory size, not the on disk size
3. use YX for chunking, not ZYX, so usual slice-by-slice in napari is much quicker (slicing by X or Y though will be then really slow, this is still better than before)
4. A new setting allowing users to control the in-memory limit for RAM

[codecov]

Codecov Report

❌ Patch coverage is 97.93814% with 2 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
src/ndevio/nimage.py	97.70%	2 Missing ⚠️

📢 Thoughts on this report? Let us know!