ops: handle multi-compute of the same weight (#13705)

If the same weight is used multiple times within the same prefetch
window, it should only apply compute state mutations once. Mark the
weight as fully resident on the first pass accordingly.

Author: rattus128

comfy/ops.py

@@ -253,6 +253,9 @@ def to_dequant(tensor, dtype):
     if bias is not None:
         bias = post_cast(s, "bias", bias, bias_dtype, prefetch["resident"], update_weight)
 
+    if prefetch["signature"] is not None:
+        prefetch["resident"] = True
+
     return weight, bias