[inductor] Insert triton barrier before storing to inplace buffers (#100769)
The linked issue demonstrates a triton bug where a load broadcasted
over multiple warps may see the result of a store that happens later
in the triton program. The workaround is to add a barrier before
storing, which enforces that all warps have already read the data.
e.g. in `test_embedding_var_mean` we now generate:
```python
tl.debug_barrier()
tl.store(in_out_ptr1 + (tl.broadcast_to(x0, [XBLOCK, 1])), tmp17, None)
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/100769
Approved by: https://github.com/jansel, https://github.com/ngimel
