Fix color when decoding image 4444

[Aireil]

I'm guessing the same issue applies to decodeImage4444split1channel and decodeImage4444split in the same file, but I couldn't find a texture using those to test, so I didn't touch them.

[goaaats]

Could you leave a comment explaining the fix, and point to it in the implementations you didn't touch?

[Aireil]

Well, not sure why it works, just saw that a status effect icon that was red in-game was blue in Explorer, so I swapped those two and it worked 🐮. I can add a comment to say that it's probably swapped in the untouched ones though yeah.

[Aireil]

There, not sure if that's what you wanted? 😳

[Charlignon]

Hello ! I ran into the same issue and independently found the same fix.
For reference, it happens on tex files 060000.win32.index/ui/icon/010000 to .../013000 (specifically status icons)

I don't think there is a need for commenting on the part that this PR fixes : the implementation was simply wrong to begin with, the LITTLE_ENDIAN byte order imply to read bgra and not argb. This seems to be a simple oversight, because other decoding methods using LITTLE_ENDIAN do use this order.

As per the split and splitChannel methods , I browsed a few thousand tex files and couldn't find any. We could blindly apply the fix or not, but in both cases I agree that there should be a comment on these methods (not on the one that was fixed though)
For the wording I suggest something like

// The channels may not be read in the correct order. If any issue occurs, check that the bytes are read in accordance to ByteOrder. You can refer to decodeImageRGBA or decodeImage4444 for other implementations.