Image Parser performed on stacks converts z to channels

Hi! I’ve been working with confocal stacks in ImageJ, and when I use the image parser on them, it produces a result that is an xyc (channel) stack instead of an xyz (z pos) stack. This makes them incompatible with downstream functions I want to use them for, and I seem to have a persistently glitchy time trying to use ‘Reorder Hyperstack’. Has anyone else run into this issue or have advice? I’ll try using some of the other calculators, but I really appreciate getting a 32-bit image out of the process as it maintains more intensity resolution.

Edit: here is a macro using one of the sample images:

run("Confocal Series (2.2MB)");
run("Split Channels");
run("Image Expression Parser (Macro)", "expression=A+B a=C2-confocal-series.tif b=C1-confocal-series.tif c=None");
run("Re-order Hyperstack ...", "channels=[Slices (z)] slices=[Channels (c)] frames=[Frames (t)]");