lang: Add more test cases related to future graph optimization

We are planning to implement an optimization in which some function
calls are compiled to a single static graph rather than to a CallFunc
which dynamically creates a sub-graph at runtime. These test cases
exercise corner cases for which it would be theoretically possible to
use a static graph, but which we might not be able to optimize.

lang/interpret_test/TestAstFunc2/constant-sub-graph.txtar

@@ -0,0 +1,17 @@
+-- main.mcl --
+import "os"
+$apply1 = func($f, $x) {
+	$f($x)
+}
+$apply2 = func($f, $x) {
+	$f($f($x))
+}
+$apply = if (os.system("echo 0; echo 1") == "0") {$apply1} else {$apply2}
+$id = func($y) { $y }
+
+# since $apply changes over time, this call needs a dynamic sub-graph. In
+# theory, the $f calls above do not need a sub-graph, but does our optimization
+# support this corner case yet?
+test $apply($id, "foo") {}
+-- OUTPUT --
+Vertex: test[foo]

lang/interpret_test/TestAstFunc2/speculative-call1.txtar

@@ -0,0 +1,11 @@
+-- main.mcl --
+$f1 = func($x) { $x }
+$f2 = func($x) { $x + $x }
+$id = func($y) { $y }
+$f = if ($id(true)) {$f1} else {$f2}
+
+# $f is always $f1, so in theory, this call does not need a sub-graph.
+# But does our optimization support this corner case yet?
+test $f("foo") {}
+-- OUTPUT --
+Vertex: test[foo]

lang/interpret_test/TestAstFunc2/speculative-call2.txtar

@@ -0,0 +1,12 @@
+-- main.mcl --
+$x = "not this x"
+$add = func($x) {
+	func($y) {$x + $y}
+}
+$addfoo = $add("foo")
+
+# making sure that $x correctly refers to the lambda parameter rather than the
+# top-level "not this x", even in the Value() codepath.
+test $addfoo("bar") {}
+-- OUTPUT --
+Vertex: test[foobar]