Prevent incorrect Tigris configuration in Rails scanner for GitHub integration

[injoongy]

Change Summary

A user reported that when using the GitHub integration to launch their app, it kept failing because it kept prompting for Tigris bucket creation and getting stuck on the terms of service agreement. Deploying/launching through flyctl was unaffected, and worked fine.

What and Why:
This was failing in the scanner/rails.go file. We had logic for checking a few different things about a user's Rails app, and if any were true, we'd automatically enable Tigris. They were:

• aws-sdk-s3 in Gemfile/Gemfile.lock
• :active_storage_attachments found in any db/migrate/*.rb
• config/storage.yml exists, then parse and look for service: S3
• sqlite3 being present in the Dockerfile and SKIP_EXTENSIONS being unset

The user sent over their Dockerfile, which had sqlite3 in it, and they also confirmed that they had old migration files that related to Active Storage. I'm fairly confident the presence of sqlite3 is what's biting them here (I'm less sure about the Active Storage stuff).

I've added changes here to try and correct these issues:

• Removed the os.Getenv("SKIP_EXTENSIONS") conditional line so that Tigris isn't enabled automatically if that env var isn't set (and I'd assume it's not set for most people).
• Added a nested conditional to only enable Tigris if Active Storage is currently not local-only.
• Updated logic to parse YAML a little better.

Screenshots:

I was able to reproduce the issue with a starter Rails app: