Store inline caches in StackHandleScope<> in HInliner.

Avoid a managed heap allocation.

Test: m test-art-host-gtest
Test: testrunner.py --host
Bug: 181943478
Change-Id: I7ce65c93ad2f59490dbfa2aaccba98b6ca1fd585
diff --git a/runtime/jit/jit_code_cache.h b/runtime/jit/jit_code_cache.h
index ea1e924..76b7f77 100644
--- a/runtime/jit/jit_code_cache.h
+++ b/runtime/jit/jit_code_cache.h
@@ -34,6 +34,7 @@
 #include "base/safe_map.h"
 #include "compilation_kind.h"
 #include "jit_memory_region.h"
+#include "profiling_info.h"
 
 namespace art {
 
@@ -305,7 +306,8 @@
       REQUIRES(!Locks::jit_lock_)
       REQUIRES_SHARED(Locks::mutator_lock_);
 
-  void CopyInlineCacheInto(const InlineCache& ic, Handle<mirror::ObjectArray<mirror::Class>> array)
+  void CopyInlineCacheInto(const InlineCache& ic,
+                           /*out*/StackHandleScope<InlineCache::kIndividualCacheSize>* classes)
       REQUIRES(!Locks::jit_lock_)
       REQUIRES_SHARED(Locks::mutator_lock_);