When applied to a theorem-tactic f and a goal ({A1,...,An} ?- C), the
EVERY_ASSUM tactical maps f over a list of ASSUMEd assumptions then
applies the resulting tactics, in sequence, to the goal:
   EVERY_ASSUM f ({A1,...,An} ?- C)
    = (f(A1 |- A1) THEN ... THEN f(An |- An)) ({A1,...,An} ?- C)