Revert "perf: buffer-at-a-time search for literal patterns"

[sylvestre]

No description provided.

[codspeed-hq]

Merging this PR will degrade performance by 94.82%

❌ 3 regressed benchmarks
✅ 7 untouched benchmarks
⏩ 17 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
❌	literal_no_match	1.3 ms	27.6 ms	-95.32%
❌	search_pattern	1.6 ms	29.5 ms	-94.56%
❌	fixed_string	1.6 ms	29.5 ms	-94.53%

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing revert-16-literal-fast-path (0b62b44) with main (7c79cb4)}

Footnotes

1. 17 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[lhecker]

We don't have to revert it outright now that it's already merged. It wasn't a wrong approach after all. (If anything, it may be not "ideal".)

Rather, I wanted to discuss an alternative approach, one that may be smaller in size. My main concern is about total code size, such that we can keep it small & maintainable, even long-term. I wanted to mention it, particularly if you were interested in further refining the approach. I wouldn't be sad if it had never happened. If you want, we can close this PR. :)