Make sure we configure the compression method and tolerance as well as the adaptive level from the primary layer when opening an existing VDS. Moved the public CompressionInfo class to the internal implementation since it is not required to be part of the public API. Re-output the Python API. Make sure the adaptive level is correctly propagated all the way to the iomanager. Added simple tests for adaptive reads from file and object store, but it doesn't actually validate that less data is read since I don't know how to do that.