Speed up uncompressed DDS reader

[akx]

Follows up on #7589. I noticed this looking at pytest --durations; a fuzzer test that just loads a DDS was surprisingly slow, and turns out doing 1-byte reads is not very efficient...

Locally:

$ pytest-benchmark compare --between=min .benchmarks/Darwin-CPython-3.14-64bit/*.json
Name (time in us)                   0001_orig Min  0002_new Min      ΔMin
-------------------------------------------------------------------------
test_load[uncompressed_rgb.dds]      405,475.7920    1,589.0000    -99.6%
-------------------------------------------------------------------------

or 255.17 times lower minimum, or in average OPS terms, +22351.7% faster. 😂

[radarhere]

len(src) // bytecount might be less if the image is truncated, but why would pixel_count ever be less? Why not always use len(src) // bytecount?