Update on "[Executorch][LLM] Use caching allocator for runner"

We observed that on iOS it improves perf by 6% because SDPA op does temp allocations.

No significant difference on android though.

Differential Revision: [D86120038](https://our.internmc.facebook.com/intern/diff/D86120038/)

[ghstack-poisoned]

Author: kimishpatel

extension/memory_allocator/cpu_caching_malloc_allocator.cpp

@@ -19,7 +19,8 @@ void* CPUCachingAllocator::allocate(size_t size, size_t alignment) {
     return nullptr;
   }
   alignment = std::max(alignment, kCachingAllocatorDefaultAlignment);
-  size_t adjusted_size = executorch::extension::utils::get_aligned_size(size, alignment);
+  size_t adjusted_size =
+      executorch::extension::utils::get_aligned_size(size, alignment);
   if (adjusted_size == 0) {
     return nullptr;
   }

extension/memory_allocator/memory_allocator_utils.h

@@ -27,8 +27,7 @@ inline size_t get_aligned_size(size_t size, size_t alignment) {
     // this is the only portable way to get aligned memory from the heap.
     const size_t extra = alignment - 1;
     if ET_UNLIKELY (extra >= SIZE_MAX - size) {
-      ET_LOG(
-          Error, "Malloc size overflow: size=%zu + extra=%zu", size, extra);
+      ET_LOG(Error, "Malloc size overflow: size=%zu + extra=%zu", size, extra);
       return 0;
     }
     size += extra;