Simplify encode_to_slice by using chunks_exact_mut

[mina86]

Rather than using generate_iter to iterate over each 2-element chunks
of output buffer, use chunks_exact_mut. This removes the need for the
custom generate_iter function and makes the code shorter.

Criterian shows improvement but to be honest I don’t trust those
results (partially because also decoding showed improvements even
though I didn’t touch it):

hex_encode              time:   [79.268 µs 81.856 µs 84.279 µs]
                        change: [-15.070% -12.678% -10.255%] (p = 0.00 < 0.05)
                        Performance has improved.

rustc_hex_encode        time:   [36.926 µs 36.949 µs 36.971 µs]
                        change: [-3.3521% -3.2536% -3.1583%] (p = 0.00 < 0.05)
                        Performance has improved.

faster_hex_encode       time:   [820.07 ns 821.16 ns 822.52 ns]
                        change: [-3.7706% -3.4712% -3.1944%] (p = 0.00 < 0.05)
                        Performance has improved.

faster_hex_encode_fallback
                        time:   [11.554 µs 11.557 µs 11.562 µs]
                        change: [-2.8788% -2.8220% -2.7676%] (p = 0.00 < 0.05)
                        Performance has improved.

hex_decode              time:   [100.90 µs 100.98 µs 101.08 µs]
                        change: [-3.9865% -3.5655% -3.2365%] (p = 0.00 < 0.05)
                        Performance has improved.

rustc_hex_decode        time:   [65.759 µs 65.788 µs 65.819 µs]
                        change: [-2.1181% -1.9775% -1.8495%] (p = 0.00 < 0.05)
                        Performance has improved.

faster_hex_decode       time:   [2.2675 µs 2.2684 µs 2.2693 µs]
                        change: [-2.9682% -2.8188% -2.6234%] (p = 0.00 < 0.05)
                        Performance has improved.

faster_hex_decode_unchecked
                        time:   [928.35 ns 929.18 ns 930.05 ns]
                        change: [-0.5734% -0.4510% -0.3314%] (p = 0.00 < 0.05)
                        Change within noise threshold.

faster_hex_decode_fallback
                        time:   [11.416 µs 11.426 µs 11.438 µs]
                        change: [-3.8020% -3.6669% -3.5267%] (p = 0.00 < 0.05)
                        Performance has improved.

[mina86]

@KokaKiwi, friendly ping.