max / makenotwork

Take the loop back out of the feed pager's arm The whole diff is a `Loop` wrapper coming off one `Arms` body, with the five ops inside it unchanged byte for byte. That wrapper was the bug: a pager's jump is one member, the loop over jumps runs a single pass while staging, so the loop's cover and the `here` arm's cover were the same bytes and the arm adopted the loop as a child. quasi 12efc11 stopped that; this is the residual without it. Regenerated and verified on astra, where the suite links: 25 of 25 residual tests pass, `every_committed_residual_matches_a_fresh_one` among them, which is what says this file is exactly what today's derivation produces.