Optimize Compressor::rebuild_from by removing full pass on the two-bytes codes

[AdamGS]

Summary

Instead of having an UNSUSED state, we start by populating the two-byte codes vec with the initial state from the 1-byte codes array, then just override it where applicable.

Impact

For the new benchmarks I get:

rebuild-from/empty-table
                        time:   [9.2503 µs 9.4427 µs 9.6325 µs]
                        change: [-68.678% -67.274% -66.086%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 1 outliers among 100 measurements (1.00%)
  1 (1.00%) low mild
rebuild-from/one-byte-table
                        time:   [9.1684 µs 9.2991 µs 9.4306 µs]
                        change: [-68.056% -66.830% -65.802%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 5 outliers among 100 measurements (5.00%)
  1 (1.00%) low severe
  4 (4.00%) low mild
rebuild-from/two-byte-table
                        time:   [9.0790 µs 9.2371 µs 9.4144 µs]
                        change: [-68.339% -67.150% -65.978%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 2 outliers among 100 measurements (2.00%)
  2 (2.00%) low mild
rebuild-from/mixed-length-table
                        time:   [9.3243 µs 9.4667 µs 9.6103 µs]
                        change: [-67.862% -66.675% -65.387%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 1 outliers among 100 measurements (1.00%)
  1 (1.00%) low mild
rebuild-from/trained-table
                        time:   [9.2450 µs 9.3939 µs 9.5343 µs]
                        change: [-67.732% -66.724% -65.681%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 1 outliers among 100 measurements (1.00%)
  1 (1.00%) low mild

[codspeed-hq]

Merging this PR will improve performance by 13.03%

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 3 improved benchmarks
✅ 22 untouched benchmarks
🆕 5 new benchmarks

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	Simulation	decompress-into-reuse	687.8 ns	600.3 ns	+14.58%
⚡	Simulation	compress-twobytes	238.6 ns	209.4 ns	+13.93%
⚡	Simulation	compress-hashtab	303.6 ns	274.4 ns	+10.63%
🆕	Simulation	mixed-length-table	N/A	91.1 µs	N/A
🆕	Simulation	trained-table	N/A	91.8 µs	N/A
🆕	Simulation	empty-table	N/A	90.6 µs	N/A
🆕	Simulation	one-byte-table	N/A	90.6 µs	N/A
🆕	Simulation	two-byte-table	N/A	90.7 µs	N/A

Tip

Curious why this is faster? Comment @codspeedbot explain why this is faster on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing adamg/optimize-rebuild-from (579bf95) with develop (fbe1ce2)}