swap_ranges benchmark

[AlexGuteniev]

-----------------------------------------------------------
Benchmark                 Time             CPU   Iterations
-----------------------------------------------------------
bm<uint8_t>/1          1.95 ns        0.854 ns    640000000
bm<uint8_t>/5          2.96 ns         1.44 ns    640000000
bm<uint8_t>/15         4.40 ns         2.00 ns    320000000
bm<uint8_t>/26         2.93 ns         1.26 ns    448000000
bm<uint8_t>/38         3.90 ns         1.59 ns    373333333
bm<uint8_t>/60         4.21 ns         1.96 ns    446008889
bm<uint8_t>/125        5.31 ns         2.51 ns    373333333
bm<uint8_t>/800        15.5 ns         6.70 ns     95573333
bm<uint8_t>/3000       75.1 ns         36.8 ns     18666667
bm<uint8_t>/9000        140 ns         69.8 ns     11200000

I actually tried to improve by using masks, but it didn't become better with masks this time.

[StephanTLavavej]

I'm mirroring this to the MSVC-internal repo - please notify me if any further changes are pushed.

[StephanTLavavej]

Thanks for monitoring the performance of this algorithm! 🏃 ⏱️ ✅