t5351: avoid using `test_cmp` for binary data

t5351: avoid using `test_cmp` for binary data

The `test_cmp` function is meant to provide nicer output than `cmp` when
expected and actual output of Git commands disagree. The implicit
assumption is that the output is line-based and human readable.

However, aaf81223f48 (unpack-objects: use stream_loose_object() to
unpack large objects, 2022-06-11) introduced a call that compares the
contents of pack files, which are distinctly not line-based nor human
readable.

This causes problems because on Windows, we hand off to the Bash
function `mingw_test_cmp` that compares the lines while ignoring line
ending differences. And this Bash function spends an insane amount of
cycles trying to read in that binary pack file, so that it is almost
indistinguishable from an infinite loop.

For example, t5351 took 1486 seconds in the CI run at
https://github.com/git/git/runs/7398490747?check_suite_focus=true#step:5:171,
to complete. And yes, that is almost half an hour.

Since Git's tests already use `cmp` consistently when comparing pack
files, let's change this instance to use `cmp` instead of `test_cmp`,
too, and fix that performance problem.

Now t5351 takes all of 22 seconds.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>