pub fn expm1<F>(x: F) -> Fwhere F: Float,
e^x - 1, without the loss of digits near 0 that exp(x) - 1 has.
exp(x) - 1