Make the matches field smaller to improve cache locality

[brian-pane]

No description provided.

[brian-pane]

On my test system, this helps a little at compression level 1, with no statistically significant change at higher levels.

Benchmark 1 (65 runs): ./blogpost-compress-baseline-native 1 rs silesia-small.tar
  measurement          mean ± σ            min … max           outliers         delta
  wall_time          77.1ms ± 1.29ms    74.5ms … 81.4ms          1 ( 2%)        0%
  peak_rss           26.7MB ± 69.9KB    26.5MB … 26.7MB          0 ( 0%)        0%
  cpu_cycles          284M  ±  592K      283M  …  286M           1 ( 2%)        0%
  instructions        536M  ±  282       536M  …  536M           0 ( 0%)        0%
  cache_references    270K  ± 16.8K      264K  …  377K           9 (14%)        0%
  cache_misses        243K  ± 3.36K      230K  …  248K          10 (15%)        0%
  branch_misses      3.07M  ± 5.14K     3.05M  … 3.08M           0 ( 0%)        0%
Benchmark 2 (66 runs): ./target/release/examples/blogpost-compress 1 rs silesia-small.tar
  measurement          mean ± σ            min … max           outliers         delta
  wall_time          75.9ms ±  889us    74.2ms … 78.6ms          1 ( 2%)        ⚡-  1.6% ±  0.5%
  peak_rss           26.7MB ± 73.5KB    26.5MB … 26.9MB          0 ( 0%)          +  0.0% ±  0.1%
  cpu_cycles          280M  ±  837K      279M  …  285M           1 ( 2%)        ⚡-  1.3% ±  0.1%
  instructions        536M  ±  246       536M  …  536M           0 ( 0%)          +  0.0% ±  0.0%
  cache_references    268K  ± 6.94K      264K  …  309K           5 ( 8%)          -  0.9% ±  1.6%
  cache_misses        244K  ± 2.27K      235K  …  247K           7 (11%)          +  0.4% ±  0.4%
  branch_misses      2.99M  ± 6.06K     2.98M  … 3.01M           2 ( 3%)        ⚡-  2.6% ±  0.1%

[folkertdev]

Cool, I don't see a change locally but overall I think this is a good change, with the possibility of shrinking the structure further in the future.