Fix: gate FrameType::isRawPointer on HotSpot

The raw-pointer bci encoding (bit 30) only ever means "jmethodID slot
replaced by a VMMethod*" on HotSpot. On OpenJ9, an unencoded, VM-supplied
bci can coincidentally have bit 30 set, which was misread as our
raw-pointer encoding and routed into JVMSupport::resolve(), which
asserts/aborts for non-HotSpot VMs. Reproduced on Semeru 8.0.462/OpenJ9
via GetLineNumberTableLeakTest.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

Author: jbachorik

ddprof-lib/src/main/cpp/frame.h

@@ -79,8 +79,11 @@ class FrameType {
     return (FrameTypeId)((bci >> TYPE_SHIFT) & FRAME_TYPE_MASK);
   }
 
+  // Raw pointers only exist on HotSpot (see hotspotSupport.cpp fillFrameRaw).
+  // On other VMs an unencoded, VM-supplied bci can coincidentally have bit 30
+  // set; that must not be mistaken for our raw-pointer encoding.
   static inline bool isRawPointer(int bci) {
-    return bci > 0 && (bci & RAW_POINTER_MASK) != 0;
+    return VM::isHotspot() && bci > 0 && (bci & RAW_POINTER_MASK) != 0;
   }
 };