[v6.d REVIEW] Unfudge min/max flattening tests

The skip-fudge expects no flattening to happen with single-arg
`[min]`/`[max]` with Range arg, but I don't see what would be a
viable answer in such a case, other than min/maxing the Range's
endpoints, which would merely be *an optimization*, not a functional
difference, and IMO should not be specced. (N.B.: don't see
an easy way to perform that opt in Rakudo, as &METAOP_REDUCE_LIST
makes a routine with a +slurpy).

Change the fudge to more-than-one arg tests, to ensure flattening
is not done there, as in these cases the result would be different
if flattening were done.

S32-list/minmax.t

@@ -133,9 +133,8 @@ is (-Inf, Inf).max, Inf,"Inf is greater than -Inf";
 #is (0, NaN).max, NaN,    "max(0,NaN)=NaN";
 #is (Inf, NaN).max, NaN,    "max(Inf,NaN)=NaN";
 
-#?rakudo 2 skip "post-GLR"
-is ([min] 42..4200000000000000000), 42, '[min] does not flatten';
-is ([max] -4200000000000000000..42), 42, '[max] does not flatten';
+is ([min]  42..420000000000000, -700), -700, 'multi-arg [min] does not flatten';
+is ([max]  42..420000000000000,  700),  700, 'multi-arg [max] does not flatten';
 
 is ([min] 5,10,-15,20), -15, 'reduce min int';
 is ([max] 5,10,-15,20), 20, 'reduce max int';