Add synchronize in aug_fwd

Author: michel2323

Diff for: ext/EnzymeExt.jl

@@ -236,9 +236,8 @@ module EnzymeExt
         TapeType, subtape, aug_kernel = _create_tape_kernel(
             kernel, ModifiedBetween, FT, ctxTy, ndrange, iterspace, args2...
         )
-        @show TapeType
         aug_kernel(f, ModifiedBetween, subtape, Val(TapeType), args2...; ndrange, workgroupsize)
-        # KernelAbstractions.synchronize(backend(kernel))
+        KernelAbstractions.synchronize(backend(kernel))
 
         # TODO the fact that ctxTy is type unstable means this is all type unstable.
         # Since custom rules require a fixed return type, explicitly cast to Any, rather