Benchmark for C++ PDQ index

[16BitNarwhal]

Summary

Query benchmark program for pdq/cpp/index/mih.h written in C++ (located pdq/cpp/bin/benchmark-mih.cpp)
Benchmark results in pdq/cpp/index/README.md

To run:

cd pdq/cpp
make bin/benchmark-mih.cpp
./benchmark-mih

Sample results (ran on Ubuntu 24.04.1 LTS, Intel Core i7-14700KF with 20 cores, 28 threads, 64GB RAM):

$ ./benchmark-mih
BRUTE-FORCE QUERY:
NEEDLE COUNT:               1000
HAYSTACK COUNT:             11000
TOTAL MATCH COUNT:          1000
SECONDS:                    0.311571
SECONDS PER MATCH:          0.000312

MIH QUERY:
NEEDLE COUNT:               1000
HAYSTACK COUNT:             11000
TOTAL MATCH COUNT:          1000
SECONDS:                    0.237358
SECONDS PER MATCH:          0.000237

[ianwal]

Suggested change

	std::chrono::time_point<std::chrono::system_clock> t1, t2;
	std::chrono::time_point<std::chrono::steady_clock> t1, t2;

Monotonic clocks should be used for measuring time intervals. system_clock is not necessarily monotonic, but steady_clock is.

[ianwal]

I'm not a moderator or official reviewer. I just saw this PR and want to help others avoid some common mistakes with std::chrono that have bitten me personally. Please defer to the moderator any suggestions.