Is SwiftSyntax actually faster?

New Issue Checklist

• Updated SwiftLint to the latest version
• I searched for existing GitHub issues

We've been migrating rules to SwiftSyntax and oss-check usually reports that they're faster now, so I was curious about how much faster main is.

hyperfine --warmup 1 "swiftlint --quiet --no-cache" "portable_swiftlint/swiftlint --quiet --no-cache"
Benchmark 1: swiftlint --quiet --no-cache
  Time (mean ± σ):      1.384 s ±  0.007 s    [User: 6.289 s, System: 0.208 s]
  Range (min … max):    1.371 s …  1.395 s    10 runs
 
Benchmark 2: portable_swiftlint/swiftlint --quiet --no-cache
  Time (mean ± σ):      1.511 s ±  0.015 s    [User: 7.253 s, System: 0.198 s]
  Range (min … max):    1.490 s …  1.529 s    10 runs
 
Summary
  'swiftlint --quiet --no-cache' ran
    1.09 ± 0.01 times faster than 'portable_swiftlint/swiftlint --quiet --no-cache'

• portable_swiftlint was built using make portable_zip on main.
• swiftlint was installed with Homebrew and is `0.49.1
• I run this on SwiftLint's repo

So 0.49.1 is ~10% faster than main. What am I missing? Maybe the SwiftLint repo is not a good example?

Start by reproducing the reported hyperfine comparison between swiftlint --quiet --no-cache and portable_swiftlint/swiftlint --quiet --no-cache on the SwiftLint repository. Review how make portable_zip builds the portable binary and compare it with Homebrew SwiftLint 0.49.1; done means explaining the roughly 10% performance difference or identifying the benchmark issue.