Replicate zarr structure / MIP issue for search first

[jluethi]

I've not fully managed to trace this. But when processing the /data/active/fractal/Liberali/1_well_15_fields_20_planes_SF_w_errors/D10_R1/220304_172545_220304_175557 dataset (see #8 (comment)) and doing MIPs, browsing the MIPs in the napari viewer has a weird behavior.
The MIPs load at low resolution, but when one zooms in, everything goes black.

I can't reproduce this with the 3D data of the same dataset, so my suspicion is that something either in replicate zarr file or MIP creation goes off. I.e. do we get MIPs for the empty.

Not sure if it's related to the . separator being used for MIPs, we should retest after #112. But the 23 well dataset that I just parsed also has MIPs with . separator and there the MIPs load without issues in the napari viewer

[jluethi]

Actually, I can reproduce this in the 23 well dataset as well when I zoom in enough! So I think this may all just be a consequence of #112

[jluethi]

With 0.1.9, this issue is solved. It was indeed an issue with the . separator, not with search first.