Potential for reducing number of ri files

Is there currently a way to generate .ri files in a more bundled way to reduce the number of files generated? Or it would seem this would be an optimization worth discussing given some statistical analysis.

1. .ri (and .rb) files are disproportionately populous compared to other file types in a typical Homebrew installation.

cd /opt/homebrew
find . -type f | sed 's/.*\.//' | sort | uniq -c | sort -nr | head

80487 ri
10922 rb
 7478 py
 6551 h
 5838 3
 2476 html
 2338 js
 1765 po
 1729 gz
 1641 elc

2. 95% of files are under 1k.

  Dir['/opt/homebrew/lib/ruby/gems/4.0.0/doc/*'].each do
    files = Dir["#{it}/**/*.ri"].select { File.file?(it) }
    files.each { puts Math.log(File.size(it), 10).floor }
  end

ri.sh | sort | uniq -c | sort -nr

66580 2
 3898 3
  193 4
    8 5

3. Some gems like RuboCop, Parser and Prism have disproportionately many files.

  Dir['/opt/homebrew/lib/ruby/gems/4.0.0/doc/*'].each do
    files = Dir["#{it}/**/*.ri"].select { File.file?(it) }
    total = files.sum { File.size(it) } # .tap { warn Math.log(it, 10).floor }
    printf "%d\t%d\t%s\n", total / 1e3, files.length, it.split('/')[8..].join('/')
  end

ri.sh | sort -n

5068	8957	rubocop-1.90.0
2929	8861	parser-3.3.12.0
2544	5406	prism-1.9.0
1558	2653	language_server-protocol-3.17.0.6
1276	1539	rdoc-8.0.0
1231	2284	yard-0.9.45
1005	2110	rubygems-4.0.20
 922	2463	rbs-4.2.0
 914	1347	concurrent-ruby-1.3.8
 888	1579	simplecov-1.2.0

No source files, tests, or entry points are named. Start by tracing how RDoc generates .ri files and compare the observed file counts and sizes; done means establishing whether a bundled representation is feasible and defining the scope of an implementation or follow-up proposal.