datalog: comparison ops require same-type operands

Type-mixed comparisons were silently inconsistent:

  <("hello", 5)  =>  no result, no error  (silent false)
  <(a, 5)        =>  raises "Expected number, got symbol"

Both should fail loudly with a comprehensible message. Added
dl-compare-typeok?: <, <=, >, >= now require both operands to share
a primitive type (both numbers or both strings) and raise a clear
"comparison <op> requires same-type operands" error otherwise.

`!=` is exempted because it's the polymorphic inequality test
built on dl-tuple-equal? — cross-type pairs are legitimately unequal
and the existing semantics for that case match user intuition.

2 new regression tests; conformance 267/267.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

lib/datalog/tests/builtins.sx

@@ -249,7 +249,29 @@
         (dl-bt-throws?
           (fn ()
             (dl-eval "p(10). q(R) :- p(X), is(R, /(X, 0))." "?- q(R).")))
-        true))))
+        true)
+
+      ;; Comparison ops `<`, `<=`, `>`, `>=` require both operands to
+      ;; have the same primitive type. Cross-type comparisons used to
+      ;; silently return false (for some pairs) or raise a confusing
+      ;; host-level error (for others) — now they all raise with a
+      ;; message that names the offending values.
+      (dl-bt-test!
+        "comparison — string vs number raises"
+        (dl-bt-throws?
+          (fn ()
+            (dl-eval "p(\"hello\"). q(X) :- p(X), <(X, 5)." "?- q(X).")))
+        true)
+
+      ;; `!=` is the exception — it's a polymorphic inequality test
+      ;; (uses dl-tuple-equal? underneath) so cross-type pairs are
+      ;; legitimate (and trivially unequal).
+      (dl-bt-test-set! "!= works across types"
+        (dl-query
+          (dl-program
+            "p(1). p(\"1\"). q(X) :- p(X), !=(X, 1).")
+          (quote (q X)))
+        (list {:X "1"})))))
 
 (define
   dl-builtins-tests-run!