[Relay] fix: add compute tag for trilu

[ganler]

fixes #13030

cc @masahi @jwfromm @junrushao

[tvm-bot]

Thanks for contributing to TVM! Please refer to the contributing guidelines https://tvm.apache.org/docs/contribute/ for useful information and tips. Please request code reviews from Reviewers by @-ing them in a comment.

• No users to tag found in teams: relay _{See #10317 for details}
• Built docs for commit e728d52 can be found here.

_{Generated by tvm-bot}

[shingjan]

If the test is passing then I guess it is fine. Just wonder if we need to add trilu here as an injective op.

[ganler]

@shingjan Thanks.

Adding that seems to bring a new error:

"""
  File "/home/jiawei/dev/tvm-pr/python/tvm/relay/op/__init__.py", line 54, in <module>
    from . import _transform
  File "/home/jiawei/dev/tvm-pr/python/tvm/relay/op/_transform.py", line 56, in <module>
    _reg.register_injective_schedule("trilu")
  File "/home/jiawei/dev/tvm-pr/python/tvm/relay/op/op.py", line 279, in register_injective_schedule
    return register_schedule(op_name, _schedule_injective, level)
  File "/home/jiawei/dev/tvm-pr/python/tvm/relay/op/op.py", line 264, in register_schedule
    fstrategy = _create_fstrategy_from_schedule(op_name, schedule)
  File "/home/jiawei/dev/tvm-pr/python/tvm/relay/op/op.py", line 195, in _create_fstrategy_from_schedule
    assert compute is not None, "FTVMCompute is not registered for op %s" % op_name
AssertionError: FTVMCompute is not registered for op trilu
"""

[shingjan]

Seems like adding this as an injective schedule will requires the FTVMCompute to be added, which may not be what we want as the topi trilu is implemented in python. Looks good to me with the original change!