Faster `findmin` without LoopVectorization.jl

General Usage Performance
3

I may be missing something, but the following naive approach seem to be faster than basic_findmin, at least for vectors that are not too long:

function naive_findmin(w, N...)
    x = @fastmath foldl(min, w)
    i = findfirst(==(x), w)::Int
    x, i
end

On my machine I get

julia> N = 300; v = rand(N);

julia> @b basic_findmin($v, $N)
322.393 ns

julia> @b naive_findmin($v)
133.571 ns

EDIT: The timing for naive_findmin depends on where the the minimum is. However, it seems to be always faster that basic_findmin.