Plan for axpy for scaled_add

[aganders3]

Hello and thanks for the great library. I am really enjoying working with ndarray (and rust-numpy with pyo3 and maturin).

I am working on implementing an algorithm where I believe blas::zaxpy might offer a significant performance improvement. I found #280 but no more recent work/discussion. Is there a plan for picking this up again? I'm happy to contribute if that would help, but I'm not sure I fully understand the issue with the reverted implementation.

In the meantime I will see if I can make it work with an unsafe block in my code. Can anyone help with what I might need to put to replace scaled_add here?

for j in minj..=maxj {
    let mut ok = out.index_axis_mut(Axis(0), j);
    let wj = kb(j as f64 - fj);
    for k in mink..=maxk {
        let mut oj = ok.index_axis_mut(Axis(0), k);
         let wk = kb(k as f64 - fk);
         let w = Complex64 {
             re: wate * wj * wk,
             im: 0.0,
         };
         oj.scaled_add(w, &dat);
}