perf: improve rmul/rdiv/rsqrt operations

[loyd]

Replace custom i256/u256 impl with the i256 crate.

i9-14900K:

group                                      new                                    master
-----                                      ---                                    ------
F128p18/rdiv (~1e5/~1e4, Ceil)             1.00     10.7±0.15ns        ? ?/sec    6.02     64.3±0.15ns        ? ?/sec
F128p18/rdiv (~1e5/~1e4, Floor)            1.00     10.4±0.15ns        ? ?/sec    6.17     64.2±0.22ns        ? ?/sec
F128p18/rdiv (~1e5/~1e4, Nearest)          1.00     11.2±0.16ns        ? ?/sec    5.75     64.6±0.16ns        ? ?/sec
F128p18/rmul (~1e4, Ceil)                  1.00      7.0±0.04ns        ? ?/sec    7.71     54.4±0.36ns        ? ?/sec
F128p18/rmul (~1e4, Floor)                 1.00      7.0±0.02ns        ? ?/sec    7.70     54.3±0.34ns        ? ?/sec
F128p18/rmul (~1e4, Nearest)               1.00      7.2±0.06ns        ? ?/sec    7.60     54.4±0.19ns        ? ?/sec
F128p18/rsqrt (MAX, Ceil)                  1.00     40.2±0.28ns        ? ?/sec    13.43   540.0±9.43ns        ? ?/sec
F128p18/rsqrt (MAX, Floor)                 1.00     39.3±0.27ns        ? ?/sec    13.35   524.8±1.47ns        ? ?/sec
F128p18/rsqrt (MAX, Nearest)               1.00     41.4±0.38ns        ? ?/sec    13.17  545.4±10.95ns        ? ?/sec
F128p18/rsqrt (adaptive, Ceil)             1.00     50.0±0.42ns        ? ?/sec    10.61   530.8±2.11ns        ? ?/sec
F128p18/rsqrt (adaptive, Floor)            1.00     49.2±0.42ns        ? ?/sec    10.55   519.0±1.83ns        ? ?/sec
F128p18/rsqrt (adaptive, Nearest)          1.00     50.6±0.38ns        ? ?/sec    10.83   547.8±2.17ns        ? ?/sec
F128p18/rsqrt (~1e4, Ceil)                 1.00     40.0±0.24ns        ? ?/sec    5.91    236.3±0.59ns        ? ?/sec
F128p18/rsqrt (~1e4, Floor)                1.00     39.4±0.28ns        ? ?/sec    5.68    223.9±3.09ns        ? ?/sec
F128p18/rsqrt (~1e4, Nearest)              1.00     41.2±0.28ns        ? ?/sec    5.84    240.5±0.77ns        ? ?/sec
F64p9/rsqrt (MAX, Ceil)                    1.00      1.0±0.01ns        ? ?/sec    31.48    31.6±0.09ns        ? ?/sec
F64p9/rsqrt (MAX, Floor)                   1.00      1.0±0.01ns        ? ?/sec    30.12    30.2±0.29ns        ? ?/sec
F64p9/rsqrt (MAX, Nearest)                 1.00      1.0±0.01ns        ? ?/sec    33.35    33.6±0.24ns        ? ?/sec
F64p9/rsqrt (adaptive, Ceil)               1.00      5.4±0.02ns        ? ?/sec    5.69     30.5±0.10ns        ? ?/sec
F64p9/rsqrt (adaptive, Floor)              1.00      4.9±0.01ns        ? ?/sec    5.96     29.1±0.67ns        ? ?/sec
F64p9/rsqrt (adaptive, Nearest)            1.00      5.5±0.02ns        ? ?/sec    5.92     32.6±0.58ns        ? ?/sec
F64p9/rsqrt (~1e4, Ceil)                   1.00      1.0±0.02ns        ? ?/sec    13.55    13.8±0.05ns        ? ?/sec
F64p9/rsqrt (~1e4, Floor)                  1.00      1.0±0.02ns        ? ?/sec    12.47    12.7±0.03ns        ? ?/sec
F64p9/rsqrt (~1e4, Nearest)                1.00      1.0±0.03ns        ? ?/sec    15.04    15.5±0.05ns        ? ?/sec

[loyd]

Need to update and check after Alexhuszagh/i256-rs#40