inconsisty between semantics in abstract gringo and implementation about clingo HOT 3 CLOSED

Working on this also uncovered a regression in clingo-5. The program

q(1..2).
{ p(X) : q(X) } :- 2 { q(X) }.

produces wrong results with clingo-5 because variable X wrongly became global.

from clingo.

Working on this also uncovered a second regression in clingo-5. The program

x :- #count { 1 : p(X+Y) } >= 0, X=1, Y=a.

produces wrong results with clingo-5 because X+Y is wrongly pulled into the rule body (clingo-4 guarantees that a term always has a value).

from clingo.

Fixed with d0283b8

from clingo.