Deploy all prod pipelines and queue-pusher.

This is the first test of the single prod release, which deploys all production pipelines. The staging pipelines are now autodeployed when we merge to integration branch, and the pipelines have been running the latest code since Jan 9.

Deploy prod and batch processing updates

This deploys both production and batch NDT pipelines. It does not deploy ss or pt pipelines, as Ya is doing independent updates that are likely to affect those.

The batch pipeline has a relatively minor tweak, to force all inserts to go to templated tables instead of partitions.

The prod pipeline has two significant changes:

1. Direct bigquery inserts to the base_tables dataset.
2. Disable snapshot deltas. We will be enabling them again sometime in the future, probably 2Q2018, but at present there isn't any demand for them, and we don't know if we have the right schema to be useful.

Push unified views to prod.

Merge pull request #396 from m-lab/views

Allow multiple aliases

Push latest queue_pusher code to sandbox

badly out of date.

Batch omit deltas and increase insert buffer size

This update should NOT impact SS and PT, but we should also update those pipelines to run the same code version.

Various minor changes

Primarily changes buffer size for omit_delta configs, to 50 rows.

canary field counts in staging

ndt-staging-v0.5.1-e

Remove _ fields

Canary delta field counting

Not doing this in sandbox, because sandbox now has deltas disabled.

no deltas - use empty slice struct

ndt-sandbox-v0.5.1-c

Change to bool, change to empty slice

Change cpu utilization back to 0.85

ndt-batch-v0.5.0c

lower cpu utilization