Fix min_index assignment in corner calculations

[viivsion]

arranged into a cycle with the lowest JMh hue value at index [1]

[linux-foundation-easycla]

The committers listed above are authorized under a signed CLA.

• ✅ login: viivsion / name: viivsion (54a588f)

[CLAassistant]

All committers have signed the CLA.

[scottdyer]

This does seem a valid fix, since in the original blink code this line is indeed written as min_index = i.

Can you provide a litte background about what prompted this review?
Do you have any comments on the implication of this change? Is it fixing a particular error you encountered?

EDIT: I guess what I'm trying to ask is - I'd love to know a bit more about what brought you to this file. Did you happen to encounter a specific image artifact or interpolation error in highly saturated gradients that led you here? Because if you have an example image where this bug visibly breaks, it would help us document the real-world impact of the fix. This is a clear logic flaw that will get fixed, just trying to make sure we can communicate when and where it might matter if implementers don't patch it right away.

[viivsion]

This change has no impact on RGB images. I only noticed this code‑writing issue while studying the code, and this conditional branch is never triggered under normal RGB images.