Non-QC catalog should not include the stores that actually passed QC

[jbusecke]

The way we have the cataloging set up right now causes all written stores to be written to the noqc catalog file.
It would be better to filter those out here so that the noqc catalog only contains iids that did not actually pass the tests.

[jbusecke]

In #89 I introduced some simplified logic which is based on a single bq table (#90). This now produces three different csv's:

• A quality-controlled (tests passed, not retracted)
• A non-quality-controlled (tests failed, not retracted)
• A retracted (test whatever, retracted)

Closing this.