feat: use instrument-hook markers in walltime

[GuillaumeLagrange]

No description provided.

[codspeed-hq]

Merging this PR will degrade performance by 30.5%

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 16 improved benchmarks
❌ 40 regressed benchmarks
✅ 221 untouched benchmarks

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	WallTime	test_sum_squares_slow	307.4 µs	345.7 µs	-11.09%
❌	WallTime	test_sum_of_squares[sum_of_squares_sum_comprehension_product]	193.4 µs	198.4 µs	-2.51%
❌	WallTime	test_generate_all_combinations[5-4]	10.2 µs	10.9 µs	-6.86%
❌	WallTime	test_iir_filter_process	3 µs	3.1 µs	-3.79%
❌	WallTime	test_sum_of_squares[sum_of_squares_for_loop_product]	160.8 µs	190.3 µs	-15.51%
❌	WallTime	test_make_lowpass	5.4 µs	5.5 µs	-2.84%
❌	WallTime	test_sudoku[initial_grid0]	7.8 µs	8.3 µs	-6.45%
⚡	WallTime	test_process_creation[cat /dev/null]	5.5 ms	5.3 ms	+4.02%
❌	WallTime	test_array_alloc[10000]	20.5 µs	21 µs	-2.35%
⚡	WallTime	test_pipe_communication[10000]	12.7 ms	8.9 ms	+42.37%
❌	WallTime	test_make_peak	6.2 µs	6.8 µs	-8.56%
❌	WallTime	test_match_word_pattern[aba-GraphTreesGraph]	103.2 µs	105.4 µs	-2.04%
❌	WallTime	test_fs_write[10000]	43.7 µs	44.9 µs	-2.69%
❌	WallTime	test_threadpool_map[100000]	3.8 s	3.9 s	-2.64%
⚡	WallTime	test_recursive_fibo_20	5.8 ms	5.6 ms	+4.35%
❌	WallTime	test_sum_of_squares[sum_of_squares_sum_labmda_power]	317.4 µs	326.6 µs	-2.82%
⚡	WallTime	test_multiprocessing_map[1000]	57.9 ms	56.1 ms	+3.3%
❌	WallTime	test_fs_write[1000]	17.7 µs	18.4 µs	-3.48%
❌	WallTime	test_hostname_resolution[8.8.8.8]	7.9 µs	8.1 µs	-2.89%
❌	WallTime	test_combination_sum[candidates0-8]	10.9 µs	11.8 µs	-7.81%
...	...	...	...	...	...

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

---

_{Comparing use-benchmark-start-end-markers (ae47707) with master (2dc7398)}

[not-matthias]

LGTM overall, but we shouldn't modify the time computation.

[not-matthias]

we shouldn't move the perf_counter_ns -> it's the measuring code

We need to measure both the time AND define our markers

[not-matthias]

same here