Evaluate non-array ISeq forms (cons/concat/lazy-seq) as calls (jolt-2rx)

eval-form treated only a reader LIST (a Janet array) as a call; a runtime-built
list — a plist or lazy-seq from cons/concat/list or ~@ (list?/seq? true but
array? false) — fell through to self-eval. So (eval (cons '+ '(1 2))) returned
the list as data instead of 3, and a macro whose output contained such a subform
left it unevaluated. Add a plist?/lazy-seq? branch that coerces to the element
array via d-realize and dispatches through eval-list; an empty list self-evals.

The analyzer already punts these forms to the interpreter (analyze's :else ->
uncompilable -> interpreter fallback), so this one interpreter branch fixes the
correctness bug across the eval and macro-expansion paths; compiling them
directly (vs punting) would be a separate perf change. Verified: conformance
355/355, syntax-quote ~@ splice, list values unchanged.

src/jolt/evaluator.janet

@@ -98,4 +98,12 @@
     (if (= 0 (length form))
       @[]
       (eval-list ctx bindings form))
+    # A non-array ISeq used as a form is a CALL too (jolt-2rx): cons/concat/list
+    # and ~@ build a plist or lazy-seq (list?/seq? true, array? false) — without
+    # this they fell through to self-eval, so (eval (cons '+ '(1 2))) returned the
+    # list as data instead of 3, and macro output containing such subforms never
+    # evaluated. d-realize coerces to the element array; an empty list self-evals.
+    (or (plist? form) (lazy-seq? form))
+    (let [arr (d-realize form)]
+      (if (= 0 (length arr)) form (eval-list ctx bindings arr)))
     form)))

test/spec/iseq-call-spec.janet

@@ -0,0 +1,19 @@
+# Specification: a non-array ISeq (plist/lazy-seq, e.g. from cons/concat/list or
+# ~@) used as a FORM is evaluated as a call, not returned as self-evaluating data
+# (jolt-2rx). The interpreter only treated a reader LIST (Janet array) as a call;
+# a runtime-built list (a plist/lazy-seq table) fell through to self-eval, so
+# (eval (cons '+ '(1 2))) returned the list instead of 3. The analyzer already
+# punts such forms to the interpreter, so the fix lives in eval-form.
+(use ../support/harness)
+
+(defspec "ISeq call forms (jolt-2rx)"
+  ["eval a cons'd call"      "3"   "(eval (cons (quote +) (quote (1 2))))"]
+  ["eval a list-built call"  "6"   "(eval (list (quote +) 1 2 3))"]
+  ["eval a concat'd call"    "10"  "(eval (concat (list (quote +)) (list 1 2 3 4)))"]
+  ["nested cons'd subform"   "7"   "(eval (list (quote +) 3 (cons (quote +) (quote (1 3)))))"]
+  ["empty list self-evals"   "()"  "(eval (list))"]
+  ["macro output via cons"   "3"   "(do (defmacro mc [] (cons (quote +) (quote (1 2)))) (mc))"]
+  ["macro output via concat" "6"   "(do (defmacro mk [] (concat (list (quote +)) (list 1 2 3))) (mk))"]
+  # regressions: vectors/quoted-data are NOT calls
+  ["vector value self-evals" "[1 2 3]" "(eval (vec [1 2 3]))"]
+  ["quoted list of data"     "(quote (1 2 3))" "(quote (1 2 3))"])