Discourse seems to have fairly high velocity in terms of change and an ambitious roadmap.
To support that it needs a lot of user feedback. I think there is a clear implicit strategy to promote tests-passed because that supports early feedback on new changes.
In return the user gets free software and new features. It’s a kind of pact. I think over time this deal seems to have been proven successful.
Stable build doesn’t really help development as much so it may not be in the business interest to promote it as much (just my opinion, I don’t speak for CDCK at all).
The other issue with stable is this, and it’s even more significant:
There are usually a lot of changes between stable versions, including significant deprecations and API changes. Involvement in tests-passed as a developer, site admin or theme creator gives you a chance to tackle changes in small bite-sized pieces, instead of having a huge mountain to climb each time you hit the next stable milestone.
To support those big jumps you are going to probably need a staging site and a bunch of tests cases to walk through.
If you don’t own any customisations yourself, you might go for stable, but you are relying heavily on others over whom you may not have any strong influence to ensure that any addons you are using are sufficiently maintained for your next upgrade. You may find that some elements lose support by the time it comes to upgrade and at that point you might find yourself in a bind. You may also find the developer doesn’t support stable at all and you may have to fork and prepare a “cut” of the plugin to support your stable build. (however, there’s a good pinning system in place so it’s not a huge amount of work)
The other piece of significance in Discourse is its very unit test intensive so the test-passed branch is usually very good from a stability perspective.