35 performance close the gap to cmark 240µs → 150µs

[sc2ben]

Resolves #35

This pull request introduces significant improvements to the Markdown parser's benchmarking and parser combinator infrastructure. The main changes include a comprehensive rewrite and expansion of the performance benchmarking in both the README.md and the benchmarking script, as well as the addition of a new combinators.zig module that encapsulates reusable parser combinators and convenience wrappers.

Key changes:

Benchmarking and Documentation Improvements

• Expanded the performance section in README.md to include new benchmarks for additional Markdown parsers (cmark, cmark-gfm), added a new memory usage table, and improved formatting and reproducibility instructions. [1] [2]
• Updated the benchmarking script in flake.nix to:
  • Benchmark more tools (cmark, cmark-gfm), add memory usage (peak RSS) measurements, and split fast and slow tool benchmarking for improved accuracy.
  • Automatically update the README.md with sorted, bolded tables for both speed and memory usage, and include the input file size. [1] [2] [3]

Parser Infrastructure

• Added a new src/markdown/combinators.zig module providing reusable Mecha-based parser combinators and convenience wrappers for block-level Markdown constructs (e.g., headings, lists, blockquotes, code fences, footnotes, indented code). This module improves code organization and reusability for the Markdown parsing logic.