fixes pr1

This is a potentially weird case and decision on what to do about it. In
general, when we print out answers, we try to make sure that they look
like valid datalog. In this case, a Racket value is being used a Datalog
constant (which is legal and part of the #lang datalog embedding), so we
can't know what syntax was used to get it in the first place... so by
using a ~v we get a distinctive printing that shows it isn't datalog.

On the other hand, I could change #lang datalog so that it's environment
is empty and this program wouldn't be referring to '()

pretty.rkt

@@ -14,7 +14,9 @@
    [(? string? s)
     (text (format "~S" s))]
    [(? number? s)
-    (text (format "~S" s))]))
+    (text (format "~S" s))]
+   [x
+    (text (format "~v" x))]))
 (define (format-variable v)
   (format-datum (variable-sym v)))
 (define (format-constant c)

tests/pr/1.rkt

@@ -0,0 +1,16 @@
+#lang datalog
+
+type(z, tz).
+
+listof(tz, ez).
+
+int(ez).
+
+casepair(z, za, tza, zb, tzb).
+
+pair(TA, TE, null) :-
+    casepair(V, VA, TA, VB, TB),
+    type(V, TV),
+    listof(TV, TE).
+
+pair(A, B, C)?