make lin2lin endian-aware for big-endian systems

[EdwardBetts]

The _get_sample32 and _set_sample32 functions in pyaudioop.py were using hardcoded little-endian byte manipulation, causing test failures on big-endian architectures like s390x.

Replace manual byte operations with struct.pack_into/unpack_from using native byte order format characters ('b', 'h', 'i').

Fixes test failure: test_pyaudioop.py::test_lin2lin on s390x

[synesthesiam]

We may want to have an endian flag instead, since the audio in WAVE files is specifically little-endian.

[EdwardBetts]

We may want to have an endian flag instead, since the audio in WAVE files is specifically little-endian.

I guess you are saying the code was correct and the tests are wrong. The tests need to assume little-endian order instead of native order for the current CPU architecture.