Fixed some misbehaviours regarding question() + added #import functionality when restoring

tests/non_reg/test_sheerka_non_reg.py

@@ -1061,26 +1061,27 @@ as:
             "def concept number",
             "set_isa(one, number)",
             "def concept q from q ? as question(q)",
+            "set_auto_eval(q)",
             "def concept is_a from x is a y as isa(x,y) pre in_context(BuiltinConcepts.EVAL_QUESTION_REQUESTED)",
-            "set_is_greater_than(BuiltinConcepts.PRECEDENCE, c:is_a:, c:q:)"
+            "set_is_greater_than(BuiltinConcepts.PRECEDENCE, c:is_a:, c:q:)",
         ]
 
         sheerka = self.init_scenario(init)
         res = sheerka.evaluate_user_input("one is a number ?")  # automatically evaluated
         assert len(res) == 1
         assert res[0].status
-        assert res[0].body
+        assert res[0].body == True  # the body MUST be a boolean
 
         res = sheerka.evaluate_user_input("foo is a number ?")  # automatically evaluated
         assert len(res) == 1
         assert res[0].status
-        assert not res[0].body
+        assert res[0].body == False  # the body MUST be a boolean
 
-        # Sanity, when there is only one 'is a' concept. It's chosen regardless of the PRE condition
+        # x is a y is supposed to be a question. It cannot be used if not in a context of a question
         res = sheerka.evaluate_user_input("one is a number")
         assert len(res) == 1
-        assert res[0].status
-        assert res[0].body
+        assert not res[0].status
+        assert sheerka.isinstance(res[0].body, BuiltinConcepts.CONDITION_FAILED)
 
     def test_i_can_evaluate_source_code_with_concept(self):
         init = [