feat: performance improvements

[Princesseuh]

Various experiments we did during dev but ended up dropping while we chased conformance, needs a lot of cleaning up, though.

[codspeed-hq]

Merging this PR will improve performance by 16.4%

⚠️ 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

⚡ 9 improved benchmarks
✅ 13 untouched benchmarks
🆕 7 new benchmarks
⏩ 10 skipped benchmarks¹

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
⚡	no plugins	2.2 ms	1.7 ms	+28.64%
⚡	MDAST + HAST plugins	2.9 ms	2.4 ms	+22.22%
⚡	noop HAST plugin	3.3 ms	2.8 ms	+17.37%
⚡	MDAST + HAST plugins	2.4 ms	2.1 ms	+15.68%
⚡	HAST replaceNode keep-children (links)	5.1 ms	4.5 ms	+14.3%
⚡	mutating HAST plugin (set id on headings)	2.2 ms	1.9 ms	+13.87%
⚡	no plugins	1.6 ms	1.4 ms	+13.33%
⚡	MDAST replaceNode keep-children (links)	2.6 ms	2.4 ms	+12.21%
⚡	noop HAST plugin (all elements)	3.1 ms	2.8 ms	+10.98%
🆕	hast-text-noop (visit text nodes, no return)	N/A	2.8 ms	N/A
🆕	hast-elements-only (setProperty per element)	N/A	3.7 ms	N/A
🆕	hast-text-only (text replace per text node)	N/A	3.7 ms	N/A
🆕	mdast-text-only (text replace per text node)	N/A	3.5 ms	N/A
🆕	mdast-text-noop (visit text nodes, no return)	N/A	2.4 ms	N/A
🆕	mutating MDAST plugin (set depth on headings)	N/A	1.9 ms	N/A
🆕	parse_no_positions	N/A	721.8 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 feat/restore-perf (95325a5) with main (44e3082)}

Footnotes

1. 10 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. ↩