Avoid importing loads_function from distributed (#1220)

Closes #1219

Not sure if we need the caching behavior originally included in `loads_function`, or if `pickle.loads` is sufficient. Any idea @madsbk ?

Authors:
  - Richard (Rick) Zamora (https://github.com/rjzamora)

Approvers:
  - Mads R. B. Kristensen (https://github.com/madsbk)
  - Peter Andreas Entschev (https://github.com/pentschev)

URL: #1220

dask_cuda/proxy_object.py

@@ -19,7 +19,6 @@
 import distributed.utils
 from dask.sizeof import sizeof
 from distributed.protocol.compression import decompress
-from distributed.worker import dumps_function, loads_function
 
 from dask_cuda.disk_io import disk_read
 
@@ -85,7 +84,7 @@ def asproxy(
             subclass = ProxyObject
             subclass_serialized = None
         else:
-            subclass_serialized = dumps_function(subclass)
+            subclass_serialized = pickle.dumps(subclass)
 
         ret = subclass(
             ProxyDetail(
@@ -440,7 +439,7 @@ def __reduce__(self):
         pxy = self._pxy_get(copy=True)
         pxy.serialize(serializers=("pickle",))
         if pxy.subclass:
-            subclass = loads_function(pxy.subclass)
+            subclass = pickle.loads(pxy.subclass)
         else:
             subclass = ProxyObject
 
@@ -882,7 +881,7 @@ def obj_pxy_dask_deserialize(header, frames):
     if args["subclass"] is None:
         subclass = ProxyObject
     else:
-        subclass = loads_function(args["subclass"])
+        subclass = pickle.loads(args["subclass"])
     pxy = ProxyDetail(obj=(header["proxied-header"], frames), **args)
     if pxy.serializer == "disk":
         header, _ = pxy.obj