backend: Support graph computation that does not return an output

There are two cases where we may not have an output after computing:
 - Prompt processing where the length of the input exceeds the batch
   size
 - Internal memory management operations such as cache defrag and shift

ml/backend.go

@@ -49,7 +49,7 @@ type Context interface {
 	FromIntSlice(s []int32, shape ...int) (Tensor, error)
 
 	Forward(Tensor)
-	Compute(Tensor) Tensor
+	Compute(...Tensor)
 	Close()
 }