Need Annex C entry for hexfloat

[timsong-cpp]

The change to the pp-number production necessitated by the introduction of hexadecimal floating point literals can break valid C++14 code:

long double operator""_p(long double);
auto x = 0e1_p+0; // parsed as 0e1_p + 0 in C++14
                  // parsed as a single pp-number instead in C++17 (and so ill-formed)

[zygoloid]

There's an incompatibility with C++98 too:

#define X(a) b##a
int b0p = X(0p+0); // OK in C++98, ill-formed in C++17

[tkoeppe]

@zygoloid: Do you want me to draft the change?

[zygoloid]

@tkoeppe That'd be great, yes please!

[tkoeppe]

@zygoloid: PHAL

[tkoeppe]

This is completed and can be closed.