`guess` with source and `guess_by_source` frequently returning PlainText lexer

[depthbomb]

Describe the bug Given many code strings, using guess({ source: ... }) or guess_by_source(...) returns Rouge::Lexers::PlainText.

To Reproduce

# frozen_string_literal: true

require 'rouge'

code = <<~CODE
  {
    "s": "t",
    "i": 123,
    "b": true
  }
CODE

formatter = Rouge::Formatters::HTMLInline.new(Rouge::Themes::Github.new)
lexer = Rouge::Lexer.guess({ source: code }) # or Rouge::Lexer.guess_by_source(code)
output = formatter.format(lexer.lex(code))

puts lexer.title
puts output

Expected behavior Correctly guesses the lexer to use.

Screenshots None

System

• Windows 10 22H2 (19045.4046)
• Ruby 3.3.0 (2023-12-25 revision 5124f9ac75) [x64-mingw-ucrt]
• Rouge 4.2.0

Additional context I'm not sure if this is entirely a library bug or I'm attempting to use the code incorrectly as there wasn't much info about the usage regarding guessing by code source.