optimization: speed up Parser#siblingsCorrection

[exoego]

Part of #1049

What is the purpose of this PR?

• Fix bug
• Fix typo
• Update specs
• Add new rule
• Add new parser
• Improve or refactor rules
• Add a new core feature
• Improve or refactor core features
• Update documents
• Others

Description

This PR optimizes Parser#siblingsCorrection in two ways:

1. Replace O(n) .findLast in .siblingsCorrection with O(1) lookup
   • Labeled as sC in the below chart
2. In addition to that optimization, sortNumber is also optimized.
   • Labeled as sC + sN in the below chart

Test data: 1500+ options as mentioned in #1049

.	v4.1.0	v4.1.1	sC	sC + sN (this PR)
3rd quartile	5.00	4.50	4.44	4.21
Median	4.97	4.45	4.40	4.16
1st quartile	4.92	4.42	4.39	4.13

As the result, the entire process time on many <option>s is 6.5% shorter compared to v4.1.1.
(4.16 / 4.45 = 0.935)

Checklist

Fill out the checks for the applicable purpose.

• Fix bug
  • Add the test that can check whether resolved the issue.
• Update specs
  • Element
    • Update global attributes
    • Update content models
    • Add also ARIA that you explored ARIA in HTML Recommendation and Editor's draft
  • Attribute
    • Update global attributes
    • Update IDL attribute
• Add new rule
  • Add a test
  • Add a README document
    • Create an issue that requests translating the doc if you want.
  • Add to index
  • Add JSON schema to schema.json
  • Add to @markuplint/config-presets
    • And add to Rulesets of base presets
      • Add to a doc file
      • Add to a doc file (ja), or create an issue that requests translating the doc if you want.
  • Add to the Playground
    • Add to a config object
• Add new parser
  • Add a test
    • Do a test with some parser.
  • Add to initializer

[YusukeHirao]

@exoego Thank you!