Stop using encoding='utf-8' with msgpack

msgpack is throwing a warning that the unpacker has deprecated the use
of encoding argument. Instead the unpacker uses `raw=False`, which
does an unpack with similar behavior to the old encoding='utf-8'
mechanism.

This will reduce a lot of warning spam for projects using msgpack
such as Keystone. The change to the kwarg "raw" from "bytes_as_raw"
happned between 0.5.1 and 0.5.2. Changing the lower constraint is
required to ensure the new arguments are matching the method
signature.

Change-Id: Iadbee3ec8def9512369a415fb4603dc05d0cbc56

lower-constraints.txt

@@ -17,7 +17,7 @@ mccabe==0.2.1
 mock==2.0.0
 monotonic==0.6
 mox3==0.20.0
-msgpack==0.5.1
+msgpack==0.5.2
 netaddr==0.7.18
 netifaces==0.10.4
 os-client-config==1.28.0

oslo_serialization/msgpackutils.py

@@ -446,7 +446,7 @@ def load(fp, registry=None):
     # here is that the unpack() function (oddly) doesn't seem to take a
     # 'ext_hook' parameter..
     ext_hook = functools.partial(_unserializer, registry)
-    return msgpack.Unpacker(fp, ext_hook=ext_hook, encoding='utf-8').unpack()
+    return msgpack.Unpacker(fp, ext_hook=ext_hook, raw=False).unpack()
 
 
 def dump(obj, fp, registry=None):
@@ -484,4 +484,4 @@ def loads(s, registry=None):
     if registry is None:
         registry = default_registry
     ext_hook = functools.partial(_unserializer, registry)
-    return msgpack.unpackb(s, ext_hook=ext_hook, encoding='utf-8')
+    return msgpack.unpackb(s, ext_hook=ext_hook, raw=False)

requirements.txt

@@ -9,6 +9,6 @@
 
 pbr!=2.1.0,>=2.0.0 # Apache-2.0
 six>=1.10.0 # MIT
-msgpack>=0.5.1 # Apache-2.0
+msgpack>=0.5.2 # Apache-2.0
 oslo.utils>=3.33.0 # Apache-2.0
 pytz>=2013.6 # MIT