Use multiple threads in generalized `winding` numbers

[juliohm]

The algorithm is embarrassingly parallel, and should be trivial to run in parallel with multiple threads:

Meshes.jl/src/winding.jl

Lines 44 to 65 in 4965450

	function winding(points, mesh::Mesh)
	assertion(paramdim(mesh) == 2, "winding number only defined for surface meshes")
	(eltype(mesh) <: Triangle) || return winding(points, simplexify(mesh))
	function w(p)
	∑ = sum(1:nelements(mesh)) do i
	v = vertices(mesh[i])
	a⃗ = v[1] - p
	b⃗ = v[2] - p
	c⃗ = v[3] - p
	a = norm(a⃗)
	b = norm(b⃗)
	c = norm(c⃗)
	n = det([a⃗ b⃗ c⃗])
	d = a * b * c + (a⃗ ⋅ b⃗) * c + (b⃗ ⋅ c⃗) * a + (c⃗ ⋅ a⃗) * b
	2atan(n, d)
	end
	∑ / oftype(∑, 4π)
	end
	[w(p) for p in points]
	end

I am opening the issue in case someone wants to give it a try.