[PyTorch][QPL] Add instance_key into MOBILE_MODULE_STATS logging. (#45517)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/45517
Add unique instance_key instead of the default one into MOBILE_MODULE_STATS logging to avoid multiple events overlaps.
ghstack-source-id: 113149453
Test Plan:
Make sure that each event's start, annotate and end are having the same instancekey:
```
09-28 23:46:03.094 19349 21069 W MobileModuleQPLObserver.cpp: TESTINGTESTING onEnterRunMethod instance_key = 1123198800, method_name = forward
09-28 23:46:03.094 19349 21069 W MobileModuleQPLObserver.cpp: TESTINGTESTING onEnterRunMethod instance_key = 1123198800, model_name = bi_pytext_v10
09-28 23:46:03.094 19349 21069 W MobileModuleQPLObserver.cpp: TESTINGTESTING onEnterRunMethod instance_key = 1123198800, model_type = FBNet
09-28 23:46:03.094 19349 21069 W MobileModuleQPLObserver.cpp: TESTINGTESTING onEnterRunMethod instance_key = 1123198800, op_list_string = ["aten::__getitem__.t", "aten::__is__", "aten::__isnot__", "aten::add.Tensor", "aten::append.t", "aten::cat", "aten::contiguous", "aten::conv1d", "aten::dim", "aten::embedding", "aten::eq.int", "aten::format", "aten::len.t", "aten::max.dim", "aten::mul.Tensor", "aten::permute", "aten::relu", "aten::softmax.int", "aten::tanh", "prepacked::linear_clamp_run", "prim::RaiseException", "prim::TupleIndex", "prim::TupleUnpack", "prim::Uninitialized", "prim::unchecked_cast"]
09-28 23:46:03.181 19349 21069 W MobileModuleQPLObserver.cpp: TESTINGTESTING onExitRunMethod instance_key = 1123198800
09-28 23:46:04.183 19349 20896 W MobileModuleQPLObserver.cpp: TESTINGTESTING onEnterRunMethod instance_key = 1521608147, method_name = forward
09-28 23:46:04.184 19349 20896 W MobileModuleQPLObserver.cpp: TESTINGTESTING onEnterRunMethod instance_key = 1521608147, model_name = __torch__.Model
09-28 23:46:04.205 19349 20896 W MobileModuleQPLObserver.cpp: TESTINGTESTING onExitRunMethod instance_key = 1521608147
```
Reviewed By: iseeyuan
Differential Revision: D23985178
fbshipit-source-id: bcd5db8dc680e3cf8d12edf865377e80693cc23b