[PATCH] aarch64: Improve codegen of AdvSIMD expf family

[Wilco Dijkstra]

Hi Joana,

> Load the polynomial evaluation coefficients into 2 vectors and use lanewise MLAs.
> Also use intrinsics instead of native operations.
> expf: 3% improvement in throughput microbenchmark on Neoverse V1.
> exp2f: 5% improvement in throughput microbenchmark on Neoverse V1.
> exp10f: 13% improvement in throughput microbenchmark on Neoverse V1.
> coshf: 14% improvement in throughput microbenchmark on Neoverse V1.

OK - pushed (with slightly reformatted commit message to avoid repetition)

Cheers,
Wilco