Add memory usage report in GRPC statistic service (#88)

* Update GRPC service proto

* Fix type

* Fix type

Author: GuanLuo

protobuf/grpc_service.proto

@@ -1031,6 +1031,33 @@ message InferBatchStatistics
   StatisticDuration compute_output = 4;
 }
 
+//@@
+//@@.. cpp:var:: message MemoryUsage
+//@@
+//@@   Memory usage.
+//@@
+message MemoryUsage 
+{
+  //@@  .. cpp:var:: string type
+  //@@
+  //@@     The type of memory, the value can be "CPU", "CPU_PINNED", "GPU".
+  //@@
+  string type = 1;
+
+  //@@  .. cpp:var:: int64 id
+  //@@
+  //@@     The id of the memory, typically used with "type" to identify
+  //@@     a device that hosts the memory.
+  //@@
+  int64 id = 2;
+
+  //@@  .. cpp:var:: uint64 byte_size
+  //@@
+  //@@     The byte size of the memory.
+  //@@
+  uint64 byte_size = 3;
+}
+
 //@@
 //@@.. cpp:var:: message ModelStatistics
 //@@
@@ -1100,6 +1127,19 @@ message ModelStatistics
   //@@     compute).
   //@@
   repeated InferBatchStatistics batch_stats = 7;
+
+  //@@  .. cpp:var:: MemoryUsage memory_usage (repeated)
+  //@@    
+  //@@     The memory usage detected during model loading, which may be used to
+  //@@     estimate the memory to be released once the model is unloaded. Note
+  //@@     that the estimation is inferenced by the profiling tools and
+  //@@     framework's memory schema, therefore it is advised to perform
+  //@@     experiments to understand the scenario that the reported memory usage
+  //@@     can be relied on. As a starting point, the GPU memory usage for
+  //@@     models in ONNX Runtime backend and TensorRT backend is usually
+  //@@     aligned.
+  //@@
+  repeated MemoryUsage memory_usage = 8;
 }
 
 //@@