I was wondering the same, then I realized I didn't even comment on your last message about requirements. So:
+++ C (2.54 mm spacing for pads) is the most important, because 2.54 mm headers are cheap, ubiquitous and easy to solder. It makes little sense to design a controller for everyone to use if they have trouble getting headers easily.
++ B (peninsula), I can do without because of the way I'll use the board, but it makes sense when you consider most custom board designs and won't add much to the overall cost (board surface).
+ A (mid-mounted USB connector) would be nice to have a lower profile controller, but not essential, micro USB is not that tall anyway.
- D (USB-C), there are pros and cons, if it drives the price too high and makes the board bigger, we can forget it; otherwise, it has its advantages over micro USB so why not, but I don't think it's essential.
-- E (switch+LED-mounting pads+holes), I'm afraid it will make routing on the board much more convoluted, or we'll lose too much space and/or locations for pads.
--- F (OSH Park's design rules), does that mean that the fab used for this controller can
reliably make boards with thinner traces/spacing than OSH Park?
I suppose most people will buy the board already assembled, and that those who want to build their own variant can rework it in KiCad if they need to, so I don't think we should focus too much on OSH compatibility. But is it safe to go with traces/spacing smaller than 6 mil anyway?
Most importantly, I think we can't please everyone. So let's choose what make sense for most people and forget about marginal uses so that we can get something going soon-ish. If the controller works, as long as the whole thing is open source, people will be able to make variations that suit more specific needs anyway.