Add null checks in tensor parser to prevent SIGSEGV (pytorch#19267)

Summary:
Add two defensive checks in `tensor_parser_exec_aten.cpp`:

1. `getMemPlannedPtr()`: Validate that the `HierarchicalAllocator`
pointer is non-null before dereferencing it. A null allocator causes a
SIGSEGV when a model has memory-planned tensors but the caller provides
zero planned buffers (T266226256 — wearables Android crash at 0.0003%
hit rate).

2. `getTensorDataPtr()`: When a tensor has `allocation_info` (indicating
it expects memory-planned storage), validate that the allocator is
non-null before any code path reaches `getMemPlannedPtr()`. This is the
primary fix; the check in `getMemPlannedPtr` is defense-in-depth.

Differential Revision: D103467786

Author: rascani

runtime/executor/tensor_parser_exec_aten.cpp

@@ -41,6 +41,10 @@ ET_NODISCARD Result<void*> getMemPlannedPtr(
     const executorch_flatbuffer::AllocationDetails* allocation_info,
     size_t nbytes,
     HierarchicalAllocator* allocator) {
+  ET_CHECK_OR_RETURN_ERROR(
+      allocator != nullptr,
+      InvalidState,
+      "HierarchicalAllocator must not be null for memory-planned tensor");
   // Normal non-constant Tensor. Allocate data using mem_id and offset.
 
   // TODO(T142455629): make the allocator actually id based and not indexed
@@ -189,6 +193,13 @@ ET_NODISCARD Result<void*> getTensorDataPtr(
   const executorch_flatbuffer::AllocationDetails* allocation_info =
       s_tensor->allocation_info();
 
+  if (allocation_info != nullptr) {
+    ET_CHECK_OR_RETURN_ERROR(
+        allocator != nullptr,
+        InvalidState,
+        "HierarchicalAllocator is null but tensor has allocation_info requiring memory-planned buffers");
+  }
+
   // External tensors.
   if (s_tensor->extra_tensor_info() != nullptr &&
       s_tensor->extra_tensor_info()->location() ==