[SPIKE] Look into solution for partitioned data encoding

[ravjotbrar]

When writing partitioned data to hdfs using spark 3.1.1, the directory layout is something like:
data/dftest.parquet/col1%3D4/part1.snappy.parquet

Instead we need it to look like
data/dftest.parquet/col1=4/part1.snappy.parquet

This encoding issue only happens when using spark 3.1.1. Either need to look into fixing the incompatibility or look into decoding the directory layout in our connector.

[alexey-temnikov]

Next steps: Check behaviour on the latest Spark versions.