if a kernel uses a global var or calls another triton kernel https://github.com/pytorch/helion/pull/1150 does not yet implement. creating this issue as a note to self