pub fn f16_to_f64(bits: u16) -> f64
IEEE 754 half-precision bits -> f64. Delegates to half::f16 for proper signaling NaN, subnormal, and rounding semantics.
half::f16