add target features for the SIMD compare256 operations

[folkertdev]

important in the generic case where avx2 is not statically enabled

Benchmark 1 (53 runs): ./target/release/examples/baseline 1 rs silesia-small.tar
  measurement          mean ± σ            min … max           outliers         delta
  wall_time          95.0ms ±  720us    93.7ms … 97.2ms          1 ( 2%)        0%
  peak_rss           26.6MB ± 64.1KB    26.5MB … 26.6MB          0 ( 0%)        0%
  cpu_cycles          362M  ± 2.45M      357M  …  370M           1 ( 2%)        0%
  instructions        851M  ±  255       851M  …  851M           0 ( 0%)        0%
  cache_references   19.9M  ±  130K     19.7M  … 20.2M           0 ( 0%)        0%
  cache_misses        475K  ± 83.1K      374K  …  820K           1 ( 2%)        0%
  branch_misses      3.06M  ± 2.25K     3.06M  … 3.07M           2 ( 4%)        0%
Benchmark 2 (57 runs): ./target/release/examples/blogpost-compress 1 rs silesia-small.tar
  measurement          mean ± σ            min … max           outliers         delta
  wall_time          87.9ms ±  918us    86.8ms … 91.3ms          3 ( 5%)        ⚡-  7.5% ±  0.3%
  peak_rss           26.6MB ± 63.1KB    26.5MB … 26.6MB          0 ( 0%)          +  0.0% ±  0.1%
  cpu_cycles          330M  ± 3.22M      326M  …  343M           4 ( 7%)        ⚡-  9.0% ±  0.3%
  instructions        776M  ±  278       776M  …  776M           0 ( 0%)        ⚡-  8.9% ±  0.0%
  cache_references   19.8M  ±  145K     19.6M  … 20.3M           1 ( 2%)          -  0.4% ±  0.3%
  cache_misses        448K  ± 75.2K      349K  …  642K           0 ( 0%)          -  5.8% ±  6.3%
  branch_misses      3.06M  ± 2.53K     3.06M  … 3.07M           0 ( 0%)          -  0.1% ±  0.0%