Improve benchmarking command line

[mingxwa]

After the update, the result number becomes more stable.

Raw data from CI build:

	MSVC on Windows Server 2022 (x64)	GCC on Ubuntu 24.04 (x64)	Clang on Ubuntu 24.04 (x64)	Apple Clang on macOS 15 (ARM64)
Indirect invocation on small objects via proxy	9,754,272 ns	8,816,234 ns	8,442,648 ns	2,504,394 ns
Indirect invocation on small objects via virtual functions	34,842,900 ns	11,528,170 ns	11,614,613 ns	2,630,067 ns
Indirect invocation on large objects via proxy	14,117,789 ns	11,189,139 ns	11,605,143 ns	4,926,853 ns
Indirect invocation on large objects via virtual functions	53,516,325 ns	14,540,389 ns	13,436,771 ns	5,010,965 ns
Basic lifetime management for small objects with proxy	1,260,437 ns	656,388 ns	682,870 ns	568,324 ns
Basic lifetime management for small objects with std::unique_ptr	5,832,715 ns	3,271,340 ns	3,297,287 ns	2,231,709 ns
Basic lifetime management for small objects with std::shared_ptr (without memory pool)	7,356,122 ns	3,801,466 ns	3,659,358 ns	2,913,971 ns
Basic lifetime management for small objects with std::shared_ptr (with memory pool)	4,113,066 ns	4,515,817 ns	4,505,196 ns	1,629,097 ns
Basic lifetime management for small objects with std::any	1,890,091 ns	2,456,667 ns	2,518,811 ns	667,104 ns
Basic lifetime management for large objects with proxy (without memory pool)	8,775,308 ns	10,891,469 ns	10,462,914 ns	3,841,652 ns
Basic lifetime management for large objects with proxy (with memory pool)	4,126,878 ns	5,940,603 ns	4,631,703 ns	1,858,271 ns
Basic lifetime management for large objects with std::unique_ptr	8,428,636 ns	11,776,444 ns	10,356,230 ns	3,504,924 ns
Basic lifetime management for large objects with std::shared_ptr (both without memory pool)	9,835,728 ns	12,520,650 ns	11,313,639 ns	4,351,325 ns
Basic lifetime management for large objects with std::shared_ptr (both with memory pool)	4,526,459 ns	6,948,993 ns	5,015,226 ns	2,712,781 ns

Comparison:

	MSVC on Windows Server 2022 (x64)	GCC on Ubuntu 24.04 (x64)	Clang on Ubuntu 24.04 (x64)	Apple Clang on macOS 15 (ARM64)
Indirect invocation on small objects via proxy vs. virtual functions	🟩 proxy is about 257.2% faster	🟩 proxy is about 30.8% faster	🟩 proxy is about 37.6% faster	🟩 proxy is about 5.0% faster
Indirect invocation on large objects via proxy vs. virtual functions	🟩 proxy is about 279.1% faster	🟩 proxy is about 30.0% faster	🟩 proxy is about 15.8% faster	🟨 proxy is about 1.7% faster
Basic lifetime management for small objects with proxy vs. std::unique_ptr	🟩 proxy is about 362.8% faster	🟩 proxy is about 398.4% faster	🟩 proxy is about 382.9% faster	🟩 proxy is about 292.7% faster
Basic lifetime management for small objects with proxy vs. std::shared_ptr (without memory pool)	🟩 proxy is about 483.6% faster	🟩 proxy is about 479.1% faster	🟩 proxy is about 435.9% faster	🟩 proxy is about 412.7% faster
Basic lifetime management for small objects with proxy vs. std::shared_ptr (with memory pool)	🟩 proxy is about 226.3% faster	🟩 proxy is about 588.0% faster	🟩 proxy is about 559.7% faster	🟩 proxy is about 186.6% faster
Basic lifetime management for small objects with proxy vs. std::any	🟩 proxy is about 50.0% faster	🟩 proxy is about 274.3% faster	🟩 proxy is about 268.9% faster	🟩 proxy is about 17.4% faster
Basic lifetime management for large objects with proxy (without memory pool) vs. std::unique_ptr	🟨 proxy is about 4.0% slower	🟩 proxy is about 8.1% faster	🟨 proxy is about 1.0% slower	🟥 proxy is about 8.8% slower
Basic lifetime management for large objects with proxy (with memory pool) vs. std::unique_ptr	🟩 proxy is about 104.2% faster	🟩 proxy is about 98.2% faster	🟩 proxy is about 123.6% faster	🟩 proxy is about 88.6% faster
Basic lifetime management for large objects with proxy vs. std::shared_ptr (both without memory pool)	🟩 proxy is about 12.1% faster	🟩 proxy is about 15.0% faster	🟩 proxy is about 8.1% faster	🟩 proxy is about 13.3% faster
Basic lifetime management for large objects with proxy vs. std::shared_ptr (both with memory pool)	🟩 proxy is about 9.7% faster	🟩 proxy is about 17.0% faster	🟩 proxy is about 8.3% faster	🟩 proxy is about 46.0% faster
Basic lifetime management for large objects with proxy (without memory pool) vs. std::any	🟩 proxy is about 27.1% faster	🟨 proxy is about 0.9% faster	🟨 proxy is about 0.3% slower	🟨 proxy is about 4.2% faster
Basic lifetime management for large objects with proxy (with memory pool) vs. std::any	🟩 proxy is about 170.2% faster	🟩 proxy is about 85.1% faster	🟩 proxy is about 125.3% faster	🟩 proxy is about 115.3% faster