Fix(server): remove unnecessary configuration of FLASK_APP

[angela-tran]

I was finally able to reproduce the issue we've been seeing in our GitHub Actions Cypress test failures. That workflow starts up client and server via Docker Compose.

I had to delete all my containers and images to finally see the issue:

2023-12-20 12:35:09 + bin/init.sh
2023-12-20 12:35:09 + flask init-db
2023-12-20 12:35:09 Error: Could not import 'app'.
2023-12-20 12:35:09 
2023-12-20 12:35:09 Usage: flask [OPTIONS] COMMAND [ARGS]...
2023-12-20 12:35:09 Try 'flask --help' for help.
2023-12-20 12:35:09 
2023-12-20 12:35:09 Error: No such command 'init-db'.

I guess somewhere in some cached layer of a Docker image, there must've been something that made my server able to find the Flask app, and only by deleting everything was I able to get my Docker Compose server to break.

The reason that server was not starting up is because FLASK_APP needs to be to eligibility_server.app:app to match the change in cal-itp/eligibility-server@60c8eb2#diff-dd2c0eb6ea5cfc6c4bd4eac30934e2d5746747af48fef6da689e85b752f39557R16.

But actually, Benefits doesn't even need to set this environment variable since eligibility-server's Dockerfile already sets it.

[github-actions]

Coverage report

The coverage rate went from 90.87% to 90.87% ➡️
The branch rate is 85%.

None of the new lines are part of the tested code. Therefore, there is no coverage data about them.

[angela-tran]

The eligibility_server.app:app that's passed to Gunicorn in eligibility-server's bin/start.sh script is just to tell Gunicorn what WSGI app to run (source: Gunicorn docs).

FLASK_APP is still needed so that Flask can find the app to start (source: Flask docs).

[machikoyasuda]

Incredible debugging and sleuthing Angela 👏 👏 🗜️