`test_many_pointwise_ops_host.py` is missing from cpu host latency in nightly dashboard.

[rdspring1]

There aren't any results for test_many_pointwise_ops_host.py in the nightly dashboards. When I run it locally, it stalls after num_iters is greater than 32.

[rdspring1]

test_many_pointwise_ops_host.py times out with commit 205d192, which is the last commit with legacy bindings on November 4.

[rdspring1]

It is recursively stuck in findRefAsSubexprOf.

#386 nvfuser::ir_utils::isFunctional (v=<optimized out>) at /opt/pytorch/nvfuser/csrc/ir/utils.cpp:1422
#387 0x0000fffdc607f5b0 in __gnu_cxx::__ops::_Iter_negate<bool (*)(nvfuser::Val const*)>::operator()<__gnu_cxx::__normal_iterator<nvfuser::Val* const*, std::vector<nvfuser::Val*, std::allocator<nvfuser::Val*> > > > (__it=..., this=<synthetic pointer>) at /usr/include/c++/13/bits/predefined_ops.h:394
#388 std::__find_if<__gnu_cxx::__normal_iterator<nvfuser::Val* const*, std::vector<nvfuser::Val*, std::allocator<nvfuser::Val*> > >, __gnu_cxx::__ops::_Iter_negate<bool (*)(nvfuser::Val const*)> > (__pred=..., __last=0xfffff7e20ab0, __first=0x141ad190) at /usr/include/c++/13/bits/stl_algobase.h:2102
#389 std::__find_if_not<__gnu_cxx::__normal_iterator<nvfuser::Val* const*, std::vector<nvfuser::Val*, std::allocator<nvfuser::Val*> > >, __gnu_cxx::__ops::_Iter_pred<bool (*)(nvfuser::Val const*)> > (__pred=..., __last=..., __first=...) at /usr/include/c++/13/bits/stl_algo.h:112
#390 std::find_if_not<__gnu_cxx::__normal_iterator<nvfuser::Val* const*, std::vector<nvfuser::Val*, std::allocator<nvfuser::Val*> > >, bool (*)(nvfuser::Val const*)> (
    __pred=<optimized out>, __last=..., __first=...) at /usr/include/c++/13/bits/stl_algo.h:519
#391 std::all_of<__gnu_cxx::__normal_iterator<nvfuser::Val* const*, std::vector<nvfuser::Val*, std::allocator<nvfuser::Val*> > >, bool (*)(nvfuser::Val const*)> (
    __pred=<optimized out>, __last=0xfffff7e20ab0, __first=...) at /usr/include/c++/13/bits/stl_algo.h:459
#392 nvfuser::ir_utils::isFunctional (v=v@entry=0x13f8c810) at /opt/pytorch/nvfuser/csrc/ir/utils.cpp:1422
#393 0x0000fffdc5d7a3f4 in nvfuser::(anonymous namespace)::findRefAsSubexprOf (from=from@entry=0x12f42b20, reference=0x13f8c810, exact=exact@entry=true)
    at /opt/pytorch/nvfuser/csrc/device_lower/pass/scalar_hoist.cpp:125
#394 0x0000fffdc5d7b5a8 in nvfuser::(anonymous namespace)::findAllocPointFromDataDependency (value=0x12f42b20, exprs=std::vector of length 2, capacity 2 = {...})
    at /opt/pytorch/nvfuser/csrc/device_lower/pass/scalar_hoist.cpp:487
#395 nvfuser::(anonymous namespace)::CommonScalarInserter::maybeInsertAllocation (this=0xffffffffae30, loop=<optimized out>)
    at /opt/pytorch/nvfuser/csrc/device_lower/pass/scalar_hoist.cpp:533
#396 0x0000fffdc5d7bc1c in nvfuser::(anonymous namespace)::CommonScalarInserter::CommonScalarInserter (common_scalar_map=..., 
    exprs=std::vector of length 2, capacity 2 = {...}, this=0xffffffffae30) at /opt/pytorch/nvfuser/csrc/device_lower/pass/scalar_hoist.cpp:520