Function bessel_i1_impl 

pub fn bessel_i1_impl<P, V, const NS: usize, const NL: usize, const NF: usize, const SCALED: bool>(
    x: V,
    t: &BesselI<V::Element, NS, NL, NF>,
) -> V
where
    V: FloatVector + TranscendentalMathWithPolicy,
    P: Policy,

$I_1(x)$, or $e^{-|x|} I_1(x)$ when SCALED.

Odd in x: computed on |x| and signed at the end, so the small arm keeps its all-positive series and the large arm keeps a positive reciprocal.