ffi 1.4: regression test for shared-selector #objc_call sites

101/101 regression tests pass (+ffi-objc-call-03-selector-sharing).

Test exercises four call sites — three sharing "init" and one
"release" — to pin the multi-site / multi-selector lowering before
1.5 changes how SEL lookups are cached.

Runtime behavior: identical before and after 1.5 (all call sites
hit nil receivers; libobjc returns 0 for void). The improvement is
visible only in the emitted IR — today:

  $ ./zig-out/bin/sx ir examples/ffi-objc-call-03-selector-sharing.sx \\
      | grep -c "call ptr @sel_registerName"
  4

After 1.5 (planned): 2 — one `sel_registerName` per unique selector
string, materialized into a static `OBJC_SELECTOR_REFERENCES_<sel>`
global at module init, then loaded at each call site. Matches the
shape clang produces for `@selector(...)`. Worth re-running the
above grep after 1.5 lands as a manual sanity check.

The IR-shape snapshot harness (auto-diff of `sx ir` output) is
deferred; for now we verify by eye.

examples/ffi-objc-call-03-selector-sharing.sx

@@ -0,0 +1,41 @@
+// Phase 1 step 1.4 (PLAN-FFI.md): multiple `#objc_call` sites
+// with the same selector. Today (after 1.3) each site emits its
+// own `call @sel_registerName(<"init">)`; the actual SEL handle
+// is recomputed per call.
+//
+// 1.5 lands selector interning: one static `SEL` global per
+// unique selector string (named `OBJC_SELECTOR_REFERENCES_init`,
+// matching clang's convention) populated once via the runtime
+// at module init. Per call site becomes a single load.
+//
+// Runtime behavior is unchanged before vs. after 1.5; the
+// improvement is visible in `sx ir` only. After 1.5 we should
+// see exactly one `sel_registerName` call per unique selector
+// string in the emitted IR — verify with:
+//
+//   ./zig-out/bin/sx ir examples/ffi-objc-call-03-selector-sharing.sx \
+//       | grep -c "call ptr @sel_registerName"
+
+#import "modules/std.sx";
+#import "modules/compiler.sx";
+
+main :: () -> s32 {
+    inline if OS == .macos {
+        // Three calls, same selector, same nil receiver. Today these
+        // emit three `sel_registerName("init")` calls. After 1.5 the
+        // emit collapses to one (cached SEL global).
+        #objc_call(void)(null, "init");
+        #objc_call(void)(null, "init");
+        #objc_call(void)(null, "init");
+
+        // A different selector — should remain distinct after 1.5
+        // (one cached SEL per unique string, not per call site).
+        #objc_call(void)(null, "release");
+
+        print("ok\n");
+    }
+    inline if OS != .macos {
+        print("skipped (not macos)\n");
+    }
+    0;
+}

tests/expected/ffi-objc-call-03-selector-sharing.exit

@@ -0,0 +1 @@
+0

tests/expected/ffi-objc-call-03-selector-sharing.txt

@@ -0,0 +1 @@
+ok