Fix trailing zero bit mask.

[bostelk]

The trailing zero bit mask (0b11110111) was zeroing the incorrect bits: bit 3 instead of bits 0 - 2 or up to the number of trailing zero bits. The default bit mask (0b11111000) should now be correct.

The result is better compression on disk. A 20 tiff Zarr volume was compressed from 702 MB to 466 MB or 66% of its original size. The apparent image quality is similar, at least to my eyes.

This work is intended to address this issue: ScrollPrize/vesuvius#4

Before and After Comparison

[bostelk]

FWIW I had to make two edits to run the script:

1. Comment out a debug line (https://github.com/schillij95/scroll2zarr/blob/main/scroll_to_ome.py#L385) and
2. Offset the tiff index to: inttiffs[minz + i] on this line (https://github.com/schillij95/scroll2zarr/blob/main/scroll_to_ome.py#L327).

[schillij95]

Thank you bostelk for getting this bug found and fixed!