Fix compatibility with JetS3t for S3 Object Copy

When JetS3t does an object copy, the source path, defined as Bucket/Object in
the x-amz-copy-source header, is fully encoded (including the slash separating
Bucket and Object).

However, Cumulus was expecting this slash to be unencoded. We now decode the
whole source path first.

cumulus/cb/pycb/cumulus.py

@@ -46,6 +46,7 @@ def init_redirector(req, bucketName, objectName):
         raise ex
 
 def path_to_bucket_object(path):
+    path = urllib.unquote(path)
     if path == "/":
         return (path, None)
     # extract out the bucket name
@@ -59,8 +60,6 @@ def path_to_bucket_object(path):
         objectName = p_a[1].strip()
         if objectName == "":
             objectName = None
-        else:
-            objectName = urllib.unquote(objectName)
     return (bucketName, objectName)
 
 def createPath(headers, path):