all the "objects" in the DLT pipeline are MVs. 

To get content from a MV, spark.read.format("delta").table(<three level name to MV>. 
"pipeline compute" is marked "in beta". Set it to None, is likely to use what has been there in the current General Release version.