BUG: Categorical Series breaks when re-serializing with msgpack.

[jreback]

xref #8732

df=pd.DataFrame([[1,'John P. Doe'],[2,'Jane Dove'],[1,'John P. Doe']])
df[1]=pd.Categorical(df[1])

DataFrames work fine

In [9]: pd.read_msgpack(df.to_msgpack()).dtypes
Out[9]: 
0       int64
1    category
dtype: object

Series are breaking on re-serialization

In [10]: pd.read_msgpack(df[1].to_msgpack())
ValueError: cannot specify a dtype with a Categorical unless dtype='category'

[jreback]

cc @iorilu