I’ve been trying to convert your manual steps into a GH component, results are pretty close (though not exactly the same):
Point deviation between our two surfaces is well within tolerance though. The difference may be explained by me using more straightforward maths than the _Flow command.
I can create another component which generates ‘lofts’ of variable transverse degrees, provided all input curves are of the same degree, and all have the same number of control points. I’d rather not try to add too much smarts to this stuff right away, partly because it’s not clear to me where potential problem lie, and partly because it would be silly of me to start duplicating algorithms already used in the native Lofter.