Switch regex grammar (from POSIX2) to ECMA (for its PERL-style non-greedy ops.)

[xparq]

The default greediness (the regex always eating its way up to the end of the entire source string, if at all possible) is not always what users want, especially when matching long structured text (like code) consisting of short atoms.

And I haven't seen "greedy" mentioned even once at the extended regex reference page, while it's readily available with ECMA's, per-pattern.

And ECMA is the JS and C++ native regex grammar anyway.