Update matmul_fp8_row_meta signature (#4168)

Summary:
Pull Request resolved: #4168

X-link: facebookresearch/FBGEMM#1248

Updates the signature to matmul_fp8_row_meta to be consistent with matmul_fp8_row.

Differential Revision: D75030969

Author: njriasan

fbgemm_gpu/experimental/gemm/triton_gemm/fp8_gemm.py

@@ -1557,7 +1557,9 @@ def matmul_fp8_row_meta(
     allow_tf32: bool = True,
     fp8_fast_accum: bool = True,
     imprecise_acc: bool = False,
-    tma_persistent: bool = False,
+    tma_persistent: bool = True,
+    no_use_persistent: Optional[bool] = None,
+    use_warp_specialization: bool = False,
 ) -> torch.Tensor:
     """Shape function for torch compile."""
     M, K = a.shape