As I've said in a previous post, finally in year 11, the GCSE teacher spent a lesson on teaching the "old fashioned" method of long division - better late than never.
I have mixed views on those methods.
I think that the grid method for multiplication is good, because it's no more error prone, avoids some of the "magic" in the traditional long multiplication, and scales out to things other than integers. So you can multiply 27 x 33 and (x + 2) x (y + 7) using the same method. Multiplying polynomials together is pretty fundamental for a lot of post-16 maths and computing, and having a reliable way to do it is pretty handy. Is there a second-year computer science degree which doesn't at some point involve operations in finite fields defined by polynomials?
I think that "chunking" for long division is fantastic for integers, because it's basically standard long division with an acceptance that you might not be able to get the intermediate results off the top of your head. It mirrors how I do mental arithmetic.
But I think it's hopeless for the people it's being taught to, and agree with you they'd be better off learning "traditional" long division.
Consider 999 divided by 17. Traditional long multiplication breaks because you immediately hit 99/17, which you don't know. If you have all those mental shortcuts for multiplying by 2, 5 and 10 and if you can do addition and subtraction reliably, then chunking does roughly what I'd do in my head: well, 50 x 17 = 850, 5 x 17 = 85, 999-850-85=64, 64/17 is obviously 3 remainder something because 64/4 is 16, fiddle fiddle fiddle, answer is 67 remainder 13. The problem is that for the people it's aimed at, they're not able to do the quick multiplications by 5, nor the addition, so they would be better off accepting that they're going to need to do some trial and error and working out 99/17 by trial and improvement. And it doesn't work well for division of polynomials, either.