Extended floating-point types and standard names

Adds optional extended floating-point types float16_t, float32_t, float64_t, float128_t, bfloat16_t, their suffixes, conversion rules and <stdfloat>.