I’ve had some success but I definitely have more chewed up rings in those thinner-gauge batches than when I’m working with 22ga or thicker. Tbh I call it the cost of doing business and melt them down for doing granulation.
One thing I’ve considered to correct exactly the issues you’ve observed is placing bits of dowel both before and after the coiled completely fill the channel, and either tightening the plate down against them, or using some sort of very thin washer to manage the height of the plate so it’s at the correct height against the top of the coil but not squishing and deforming it.
As for the issue of finding dowels that fit precisely, I have a couple of ideas (though I will stipulate that these are purely theoretical and I have not tested them yet): first, sand down the dowel to fit. It’s some work, but to get a clean cut batch it should be worth it (and I think one dowel should work for multiple batches). Second, I’ve seen sets of “telescoping rods” which are actually tubes made from copper or brass and sold in hobby shops or on Amazon. They’re smaller than the selection of wooden dowels I usually see, and it seems like they come in a closer assortment too (less difference from one size to the next). They’d most likely be sacrificed for each batch, but they’re cheap and you’d get a couple batches per foot of cheap tube, so it’s not a lot to invest. You could even drill a hole in them, wind the wire directly on, and pop it right into the coil holder in one piece the whole way through.
I know it’s frustrating, but it seems to me (and I’ll always be the first to say I’m still a bit of a n00b) like a lot of this trade is always going to be trial and error even with the clearest, most detailed instructions.