Update gems blacklist for nvidia (#222)

### PR Category
Core
### PR Type
Improvements
### Description
Update gems blacklist for nvidia platform

---------

Co-authored-by: Physics <3150105638@zju.edu.cn>
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

Author: 3 people

vllm_fl/dispatch/backends/flaggems/flaggems.py

@@ -158,6 +158,7 @@ def attention_backend(self, use_mla: bool = False, use_sparse: bool = False) ->
 
         if use_sparse:
             raise ValueError("use_sparse=True requires use_mla=True.")
+        # TODO: return "vllm_fl.dispatch.backends.flaggems.impl.attention.AttentionFLBackend"
 
         return AttentionBackendEnum.TRITON_ATTN.get_path()
 

vllm_fl/dispatch/config/nvidia.yaml

@@ -45,12 +45,6 @@ op_backends:
 
 # FlagOS operator blacklist
 flagos_blacklist:
-  - copy_
-  - to_copy
-  - cat
-  - index
-  - mul
-  - pow_scalar
   - index_put_
   - index_put
   - _index_put_impl_