CI frequently failing due to long-running app tests

[conorsch]

Description

The CircleCI runs occasionally fail during the staging tests during the application test run, staging:

Too long with no output (exceeded 10m0s)

Steps to Reproduce

There's no clear way to trigger this particular failure, but here are some examples of the failure occurring:

• https://circleci.com/gh/freedomofpress/securedrop/17464
• https://circleci.com/gh/freedomofpress/securedrop/17515

Expected Behavior

Application tests run completely, reporting pass or fail.

Actual Behavior

Application test run is aborted, build is canceled, marked failed.

Comments

We can append a no_output_timeout parameter to the Provision staging servers and run tests step in the CircleCI config. See docs here. Let's bump that from 10m to 20m to hold us over until we revisit the test run strategy. Once we merge the ongoing work described in #3659, the tests run will emit output regularly, sidestepping the need for the no_output_timeout override.

[conorsch]

@emkll astutely pointed out that the new kernels are causing the builds to run longer, and the build and the test logic currently run in the same CircleCI step.