wasm: use wider loads/stores in copy_match

[folkertdev]

our baseline is d913cf5

Benchmark 1 (9 runs): wasmtime target/wasm32-wasi/release/wasm-zlib-benchmark.wasm inflate miniz_oxide
  measurement          mean ± σ            min … max           outliers         delta
  wall_time           619ms ± 3.21ms     615ms …  626ms          3 (33%)        0%
  peak_rss           82.2MB ± 2.46KB    82.2MB … 82.2MB          0 ( 0%)        0%
  cpu_cycles         2.60G  ± 6.08M     2.59G  … 2.61G           0 ( 0%)        0%
  instructions       7.05G  ±  890      7.05G  … 7.05G           0 ( 0%)        0%
  cache_references   29.0M  ±  159K     28.8M  … 29.2M           0 ( 0%)        0%
  cache_misses        895K  ± 24.3K      855K  …  925K           0 ( 0%)        0%
  branch_misses      32.7M  ± 6.52K     32.7M  … 32.7M           0 ( 0%)        0%
Benchmark 2 (13 runs): wasmtime target/wasm32-wasi/release/wasm-zlib-benchmark.wasm inflate zlib-rs
  measurement          mean ± σ            min … max           outliers         delta
  wall_time           405ms ± 2.18ms     403ms …  410ms          0 ( 0%)        ⚡- 34.6% ±  0.4%
  peak_rss           82.2MB ± 1.14KB    82.2MB … 82.2MB          1 ( 8%)          -  0.0% ±  0.0%
  cpu_cycles         1.62G  ± 8.24M     1.61G  … 1.64G           2 (15%)        ⚡- 37.8% ±  0.3%
  instructions       5.79G  ± 1.82K     5.79G  … 5.79G           1 ( 8%)        ⚡- 17.9% ±  0.0%
  cache_references   29.1M  ±  203K     28.8M  … 29.5M           0 ( 0%)          +  0.3% ±  0.6%
  cache_misses        896K  ± 19.1K      873K  …  928K           0 ( 0%)          +  0.1% ±  2.2%
  branch_misses      13.3M  ± 6.96K     13.3M  … 13.3M           0 ( 0%)        ⚡- 59.3% ±  0.0%

The wider loads and stores of this PR help quite a bit:

Benchmark 1 (13 runs): wasmtime baseline.wasm inflate zlib-rs
  measurement          mean ± σ            min … max           outliers         delta
  wall_time           409ms ± 5.14ms     405ms …  424ms          1 ( 8%)        0%
  peak_rss           82.2MB ± 1.80KB    82.2MB … 82.2MB          3 (23%)        0%
  cpu_cycles         1.62G  ± 11.2M     1.61G  … 1.66G           1 ( 8%)        0%
  instructions       5.79G  ± 1.26K     5.79G  … 5.79G           0 ( 0%)        0%
  cache_references   29.1M  ±  214K     28.9M  … 29.6M           0 ( 0%)        0%
  cache_misses        897K  ± 25.5K      867K  …  948K           0 ( 0%)        0%
  branch_misses      13.3M  ± 2.96K     13.3M  … 13.3M           0 ( 0%)        0%
Benchmark 2 (13 runs): wasmtime target/wasm32-wasi/release/wasm-zlib-benchmark.wasm inflate zlib-rs
  measurement          mean ± σ            min … max           outliers         delta
  wall_time           392ms ± 1.29ms     391ms …  395ms          0 ( 0%)        ⚡-  4.1% ±  0.7%
  peak_rss           82.1MB ± 28.3KB    82.1MB … 82.2MB          3 (23%)          -  0.0% ±  0.0%
  cpu_cycles         1.55G  ± 3.08M     1.55G  … 1.56G           0 ( 0%)        ⚡-  4.3% ±  0.4%
  instructions       5.70G  ± 1.61K     5.70G  … 5.70G           1 ( 8%)        ⚡-  1.5% ±  0.0%
  cache_references   29.3M  ±  143K     29.2M  … 29.7M           2 (15%)          +  1.0% ±  0.5%
  cache_misses        873K  ± 18.6K      850K  …  916K           0 ( 0%)          -  2.7% ±  2.0%
  branch_misses      10.6M  ± 2.34K     10.6M  … 10.6M           0 ( 0%)        ⚡- 20.5% ±  0.0%