fix: correct interpretation of extra samples with unspecified type of alpha

[soyosaki]

The commit should correct how TIFF images with samples count larger than photometric samples count are interpreted. For example, if a TIFF image claims a RGB photometric interpretation, a 8 8 8 8 bits-per-sample array, and an unspecified extra samples field (i.e. ExtraSamples = 0), the current commit should interpret the image as an RGBA image with the extra sample be treated as the alpha channel of the image, whereas currently the crate would interpret it as a RGB image and produce a corrupted pixel map.

The test extra_bits_rgb is fixed together, and it should now align with the logic of extra_bits_gray and write_extra_bits_image.

[soyosaki]

The latest commit should address the decode issue related to TIFF data in RGB with more than 1 extra samples. Clippy warning appear unrelated to the changes on the current branch.

[soyosaki]

Merged latest main, now all checks are passed.