perf+55.99%: use mimalloc for few individual utils

[oech3]

cargo build -p uu_split -p uu_ls still builds split using mimalloc which has very bad performance.
But it does not appear at CodSpeed (clean build). So this should not be merged yet.

[codspeed-hq]

Merging this PR will improve performance by 55.99%

⚡ 50 improved benchmarks
✅ 94 untouched benchmarks
⏩ 182 skipped benchmarks¹

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	Simulation	wc_bytes_synthetic[500]	194.8 µs	175.6 µs	+10.93%
⚡	Simulation	b64_decode_synthetic	171.3 µs	152.7 µs	+12.17%
⚡	Simulation	b64_encode_synthetic	167.3 µs	151.6 µs	+10.34%
⚡	Simulation	b64_decode_ignore_garbage_synthetic	168.4 µs	154.1 µs	+9.23%
⚡	Simulation	tsort_tree_dag[(10, 3)]	29.7 ms	27.7 ms	+7.09%
⚡	Simulation	tsort_linear_chain[1000000]	1.4 s	1.2 s	+14.8%
⚡	Simulation	tsort_wide_dag[100000]	117.4 ms	100.5 ms	+16.8%
⚡	Simulation	tsort_complex_dag[50000]	74 ms	65.5 ms	+13%
⚡	Simulation	cp_recursive_deep_tree[(120, 4)]	13.5 ms	12.6 ms	+7.41%
⚡	Simulation	cp_recursive_wide_tree[(6000, 800)]	188.6 ms	174.6 ms	+8.01%
⚡	Simulation	cp_preserve_metadata[(5, 4, 10)]	74.4 ms	67.1 ms	+10.81%
⚡	Simulation	cp_large_file[16]	392.7 µs	332.1 µs	+18.27%
⚡	Simulation	cp_archive_balanced_tree[(5, 4, 10)]	77.3 ms	69.9 ms	+10.57%
⚡	Simulation	cp_recursive_balanced_tree[(5, 4, 10)]	71.8 ms	65.2 ms	+10.08%
⚡	Simulation	ls_recursive_wide_tree[(10000, 1000)]	53.2 ms	41.1 ms	+29.46%
⚡	Simulation	ls_recursive_long_all_deep_tree[(100, 4)]	3.1 ms	2.5 ms	+23.68%
⚡	Simulation	ls_recursive_long_all_wide_tree[(15000, 1500)]	141.2 ms	107.8 ms	+31.03%
⚡	Simulation	ls_recursive_balanced_tree[(6, 4, 15)]	52.7 ms	33.8 ms	+55.99%
⚡	Simulation	ls_recursive_mixed_tree	1.4 ms	1.1 ms	+31.36%
⚡	Simulation	ls_recursive_long_all_mixed_tree	2.7 ms	2.2 ms	+26.95%
...	...	...	...	...	...

ℹ️ Only the first 20 benchmarks are displayed. Go to the app to view all benchmarks.

---

_{Comparing oech3:mimalloc (3c0465a) with main (5d03f63)}

Footnotes

1. 182 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[oech3]

Still cargo build -p uu_true -p uu_ls pollutes true binary...

[oech3]

cargo build -p uu_true -p uu_ls pollutes true binary

$ nm target/debug/true |grep mi_malloc
0000000000145440 T _mi_malloc_generic
0000000000138ef0 T mi_malloc
0000000000138ef0 t mi_malloc.localalias
00000000001398b4 T mi_malloc_aligned
0000000000139391 t mi_malloc_is_naturally_aligned

@naoNao89 Any idea to avoid it?