some attempts at fixing staticT; breaks lots of tests

compiler/semexprs.nim

@@ -692,20 +692,6 @@ proc evalAtCompileTime(c: PContext, n: PNode): PNode =
       else: return result
     result.typ = semfold.getIntervalType(callee.magic, call)
 
-  block maybeLabelAsStatic:
-    # XXX: temporary work-around needed for tlateboundstatic.
-    # This is certainly not correct, but it will get the job
-    # done until we have a more robust infrastructure for
-    # implicit statics.
-    if n.len > 1:
-      for i in 1 .. <n.len:
-        # see bug #2113, it's possible that n[i].typ for errornous code:
-        if n[i].typ.isNil or n[i].typ.kind != tyStatic or
-            tfUnresolved notin n[i].typ.flags:
-          break maybeLabelAsStatic
-      n.typ = newTypeWithSons(c, tyStatic, @[n.typ])
-      n.typ.flags.incl tfUnresolved
-
   # optimization pass: not necessary for correctness of the semantic pass
   if {sfNoSideEffect, sfCompileTime} * callee.flags != {} and
      {sfForward, sfImportc} * callee.flags == {} and n.typ != nil: