I am a product manager at Cloudera working on SQL-on-Hadoop. I will answer your question regarding support for Tez in CDH.
Both Spark and Tez generalize the Mapreduce distributed computing framework that underlies many tools in the Hadoop ecosystem. At Cloudera, we believe that Spark is more suitable successor of Mapreduce because of its better architecture, more features, and a bigger community with ~100 contributors from ~25 companies. Hence, the support for Spark in CDH. Supporting Tez in addition to Spark in CDH would have added a lot of complexity to the platform while creating little value for our customers. Therefore, we decided not to support Tez in CDH.
You can read more about Cloudera's views on Sparkhere. I will be happy to answer any further questions.