perf: use plain iterator for payload unmasking

[lucab]

This tweaks unmask_easy logic in order to directly use an enumerating iterator. It allows the compiler to generate more optimal machine code by performing internal iteration directly on the input slice.

[lucab]

On my Linux laptop with AMD Ryzen 7, this shows a small but reliable micro-benchmark improvement:

Finished release [optimized] target(s) in 16.03s
Running benches/unmask.rs (target/release/deps/unmask-c8256fdf566f5dbf)

unmask2/unmask 64 << 20 time:   [2.3568 ms 2.3596 ms 2.3641 ms]
                        thrpt:  [26.437 GiB/s 26.488 GiB/s 26.519 GiB/s]
                 change:
                        time:   [-8.4860% -7.2746% -6.0839%] (p = 0.00 < 0.05)
                        thrpt:  [+6.4780% +7.8453% +9.2729%]
                        Performance has improved.
Found 6 outliers among 100 measurements (6.00%)
  1 (1.00%) high mild
  5 (5.00%) high severe

[littledivy]

LGTM, thanks