The rawData flag is set on the aggregate command. For viewful timeseries queries when rawData is true we need to translate the namespace from the view to the buckets namespace, so the query can be executed on the buckets. This translation does not happen for subpipeline execution, so the subpipeline is executing like rawData is false, but the top level pipeline is executing with rawData is true. 

This is wrong, because all other options, such as allowDiskUse and collation in the aggregate command apply to the top level and subpipeline. We should do the translation from view to bucket namespace for subpipelines run on viewful timeseries.