Pixel math - Inconsistencies using median()/mean()

Siril version:

1.4.0-beta4

Operating System: Windows

Description of the bug

I'm not sure if I'm just misunderstanding the intention of how this math works with pixel math, but there appears to be inconsistencies when using mean() or median() to compare images. Here are some examples with two B&W (ha, oiii):

mean(h) =0.029376

mean(o) =0.043400

mean(h) + mean(o) =0.086801 (expected =0.072776)

mean(h) + mean(h) =0.058751 (as expected)

mean(o) - mean(o) =0.00 (as expected)

mean(o) - mean(h) =0.00 (expected =0.014024)

Similar issues are present with RGB images, and with median() as well.

Reproduction

Reproduction steps:

• Open pixel math
• Load any two images (RGB or B&W)
• Utilize both images with mean()/median() in a single expression (see examples above)

Expected result:

See examples in description

Actual result:

See examples in description

Additional information

-