Improving transaction sponsor blockspace efficiency

A little push-back here: this precludes batched CPFP, which is somewhere in-between single-tx exo fees and single CPFP exo fees, depending on how many “well I was just making txs already” cases are in play.

I suspect if we really wanted to squeeze every byte(I’m unconvinced, but this thread is for people who are interested in doing so :slight_smile: ), you’d want to support both this type of sponsor and the more general original one, with the associated complexities involved.

(or we figure out a softfork for robust single tx exo fees, with stacking ala @rustyrussell dream world!)