How to override type inference for partition columns in Hive partitioned dataset using PyArrow?

Question

I have a partition column in my Hive-style partitioned parquet dataset (written by PyArrow from Pandas Dataframe) with an entry like "TYPE=3860877578". When trying to