I think the main concern would be the gcc bugs (such as the Windows ones I previously mentioned) that might thereby be exposed.
And they'll likely point out that one can always assign using Perl_strtod via the POSIX module with POSIX::strtod() or POSIX::strtold().
Couple that with their disinterest in the matter (as clearly demonstrated by the lack of replies to
) and I think I'd have to present a much stronger case before anything was done.