Add globally accessible A/B testing opt-out setting

[tmancey]

Preferably under the Privacy and Security setting, along with P3A and the ping setting

[ShivanKaul]

Given that almost all A/B tests are going to be via P3A, I don't think it makes sense to have a separate toggle for this.

[haddadi]

most A/B product/design tests might not have any user data (will a red lion logo be more effective on shields than a blue lion logo?) in which case, there is no user data involved.

[diracdeltas]

I noted this in a DM to Hamed, but A/B test data (griffin state) is sent in crash reports too not just P3A. I think we are all on the same page that it'd be good to have a global opt out for A/B testing, so my understanding is that we can do this 3 ways:

1. Tie it to P3A opt out, and P3A opt out also opts out of sending Griffin data in crash reports. This seems a bit non-intuitive to me.
2. Explain in settings that Brave does A/B testing and to truly opt out from A/B test data collection you have to disable both the P3A and crash report flags. EDIT: see comment below, this is a lot clunkier than I initially thought.
3. Separate toggle for A/B testing. This would be cleanest. To avoid cluttering privacy settings, we can create a new section in brave://settings/privacy (which is really privacy+security) called "Your data" and put p3a, usage pings, crash reports, a/b test in there.

EDIT: Noting that https://github.com/brave/reviews/issues/1507 also collects A/B experiment data outside of P3A. Given that we'd have to make sure if opt out is tied to P3A, it also disables this data collection in ads confirmations. Or explain that to truly opt out of A/B testing you have to also opt out of rewards/news.

[ShivanKaul]

FWIW I just don't think we need to treat A/B tests as special cases of data collection.

[tmancey]

Added needs-discussion label. Thanks

[diracdeltas]

Is this actually a duplicate of #15711 ? It seems in there we had consensus to add a new toggle for opting out of "staged roll-outs, feature testing and user studies" which is even more comprehensive than this issue, which is just to opt out of user studies without actually removing the ability to get new griffin variations.

[ShivanKaul]

Yeah, that one is a full-on "no griffin for me" toggle. If we add a Data Collection subsection like is the plan in #36363, it would be fine to add a Griffin opt-out toggle there, and that should cover this case as well.