Only apply RemoveStridePadding() when bytes per pixel is one and fix #1183

[BobLd]

fix #1183

cc @rhuijben

[BobLd]

tests fail because size of of png is different between net47 / net8 / net9

[rhuijben]

Size in bytes, nor hash after recompressing is guaranteed as they have different Zlib versions. Will likely be more different in DotNet 10 as that starts using zlib-ng.

Size difference in pixels would be very surprising. I debugged the issue with .net 8 or 9)

[BobLd]

Agreed yes for the pixels, I wasn't aware for zlib. Ill have a look. Thanks again!

[rhuijben]

👍 looks good to me. You could use a checksum over the actual data instead of copying all those bytes. But I think it is hard to avoid reading it using something that really understands the image.

In theory if you can assume png you could try to locate the uncompressed data checksum... But I would recommend not going that deep in a format.

[BobLd]

@rhuijben thanks a lot for that. Agreed, as this is just for testing purpose, I don't think there's a need to improve it too much. Thanks again for finding the fix!