Bug: .{0,99} and .* give different results, even without long lines in content

[jorendorff]

Steps to reproduce:

1. Add the unit test below to onig/lib.rs, inside the tests module.
2. cargo test test_space_star_dot_star

    #[test]
    fn test_space_star_dot_star() {
        let re1 = Regex::new(r"^\s*B\n\s*M\s*.{0,99}\R").unwrap();
        let re2 = Regex::new(r"^\s*B\n\s*M\s*.*\R").unwrap();
        let content = "B\n  M;\n";
        assert!(re1.find(content).is_some()); // passes
        assert!(re2.find(content).is_some()); // fails
    }

I think both regexes should match. But re1 succeeds and re2 fails!

Doing cd onig_sys/oniguruma && git checkout master && git pull origin master did not change the behavior.

Ruby 3.0.5p211 does not have this bug. Both regexes match:

re1 = /^\s*B\n\s*M\s*.{0,99}\R/
re2 = /^\s*B\n\s*M\s*.*\R/
p re1.match?("B\n  M;\n")  # prints true
p re2.match?("B\n  M;\n")  # also prints true

[jorendorff]

Filed upstream: kkos/oniguruma#284