[libc++] Optimize rotate

[philnik777]

----------------------------------------------------------
Benchmark                                 old          new
----------------------------------------------------------
BM_std_rotate<int>/1                 0.707 ns      2.83 ns
BM_std_rotate<int>/2                 0.706 ns      2.83 ns
BM_std_rotate<int>/3                  4.93 ns      5.04 ns
BM_std_rotate<int>/4                  5.12 ns      5.14 ns
BM_std_rotate<int>/32                 41.3 ns      16.5 ns
BM_std_rotate<int>/64                 94.1 ns      42.8 ns
BM_std_rotate<int>/512                 764 ns       194 ns
BM_std_rotate<int>/4096               5899 ns      2687 ns
BM_std_rotate<int>/65536             94248 ns     43003 ns
BM_std_rotate<__int128>/1            0.713 ns      2.16 ns
BM_std_rotate<__int128>/2            0.711 ns      2.16 ns
BM_std_rotate<__int128>/3             5.63 ns      5.70 ns
BM_std_rotate<__int128>/4             5.64 ns      5.62 ns
BM_std_rotate<__int128>/32            46.1 ns      29.8 ns
BM_std_rotate<__int128>/64            98.6 ns      75.6 ns
BM_std_rotate<__int128>/512            752 ns       486 ns
BM_std_rotate<__int128>/4096          5849 ns      4913 ns
BM_std_rotate<__int128>/65536        97091 ns     78037 ns
BM_std_rotate<std::string>/1         0.722 ns     0.717 ns
BM_std_rotate<std::string>/2         0.722 ns     0.718 ns
BM_std_rotate<std::string>/3          4.35 ns      4.42 ns
BM_std_rotate<std::string>/4          6.12 ns      6.08 ns
BM_std_rotate<std::string>/32         46.7 ns      46.4 ns
BM_std_rotate<std::string>/64         95.7 ns      94.9 ns
BM_std_rotate<std::string>/512         795 ns       792 ns
BM_std_rotate<std::string>/4096       6500 ns      6442 ns
BM_std_rotate<std::string>/65536    106532 ns    103567 ns

Fixes #54949
Fixes #39644