Splits a double-precision floating-point number into a normalized fraction and an integer power of two.
