Use Prism instead of Ripper for Ruby source tokenization

[st0012]

[matzbot]

🚀 Preview deployment available at: https://a7caabcf.rdoc-6cd.pages.dev (commit: 1433d3f0854832d342f5395956f2074ed85378d2)

[tompng]

I think creating a new syntax highlighter based on Prism.lex is another option. Syntax highlighting doesn't need state bits. These are used in rdoc/parser/ruby.rb to parse ruby code. I think ripper_state_lex.rb is a bit complicated to be used as a prototype of prism based syntax highlighter