The leading idea is 3 tabs which rotate into slots.  The problem is progressive stretching of PLA, printing the tabs with supports & enough precision.  There could be horizontal tabs which screw into the existing magnet positions.  As the tabs wore out, they could be replaced.  It would eliminate the supports.  It's all a bit janky without metal.

A few printouts had problems staying in place.  The tabs aren't big enough to flex into locking notches.   A spring would be required for a rotating mechanism to stay in place.

It became clear over many prototypes that the booster can flex.  Instead of rotating, it could flex to insert the tabs into the slots.  It's just never going to be tight without springs or magnets.  There could be spring loaded ball bearings in front.

The trick with the flexible booster concept is to 1st insert the rear tab while putting the front tabs on top.  Then insert the front tabs by bending them out.  Remove the front tabs 1st by bending them out & putting them on top.  Then press back to remove the rear tab.  How long it lasts is a matter of experimentation.  It may be that the tabs can be a lot less aggressive.  It takes 13 hours to print a new rocket with different tab sizes.  A temporary adhesive might get it down to 8 hours.  The chines don't line up as well as they did with magnets.

The tab mechanism was a pain with a fully printed booster.  There's no way it would be practical with incense loaded.  The booster tabs can't be less aggressive because they stick to the bed.

Less aggressive tabs can be printed in for the thrust section without reprinting the booster.  Another idea was making the bottom tabs rotate into position, but the screws would loosen.  The screws can be temporary, since the thrust section isn't hard to print.

A new single piece engine section with temporary screws for gluing was finally designed.  It was surprisingly indistinguishable from the multi piece engine section.  The temporary screws leave holes which can be filled with hot snot.

The ship was glued without the booster so the booster could be reprinted.  The booster might actually stay attached to the grid fins without glue or with hot snot.  It stays pretty attached with friction.

A slightly modified tab section was printed to make it easier to get the booster on.  It may eventually be easier to leave the rocket permanently stacked if a bigger hole for the ashes is designed.

The correct chines were designed.