Enormous bounding boxes and cells broken into pieces

[vzinche]

Some cells have enormous bounding boxes, for example cell_ids 47 and 48 span almost the whole z-axis.
Looks like the same cells can also be split into multiple connected components (but that's hard to say in low res).

[constantinpape]

I double checked this and the bounding boxes are correct. For some reason these cells have some completely disconnected pieces at the other end of the volume. There could be multiple reasons for this, but I assume that this is due to some of the post-processing to fix segmentation errors.
I don't think it's worth fixing this now, because we would need to update the version etc. to account for the relabeling after splitting of the unconnected pieces. Gonna leave this issue open for reference.