Propose Unleash trial (for feature-flagging and split-testing) more formally via radar

[joeyAghion]

There's been a lot of interest in feature-flagging and split-testing recently. A few trials of specific tools are in intermediate states while we evaluate or unwind them, so let's make that more clear in our tech radar. Specifically:

• Trial Unleash for feature-flagging and split-testing. Support has been added to 2 back-end systems so far, and work is in progress (or planned) to integrate it into clients like Force and Eigen. Integrating into multiple clients is worthwhile despite not yet being fully "adopted," because how the service operates across the platform and stacks is an important element of this evaluation. There's some risk of wasted work, but little long-term maintenance burden because client SDKs fall back to defaults when the service is unavailable. <Debate...?>
• Avoid new LabFeatures since they're easily replaced by Unleash's built-in "user ID strategy."
• Avoid new SiteFeatures since they're easily replaced by Unleash's built-in "standard strategy."
• Document that we strive to split-test product changes in parallel, rather than simply monitoring metrics after the fact. Due to the seasonality and event-sensitive nature of our product, retroactive monitoring is hardly enough to confirm that product improvements are actually that.

The interactive version of these changes can be viewed here.

Q: Should we document Split.io's "hold" status here, or is that already obvious enough to anyone involved?

Note: other diffs are automated and unrelated.

[kathytafel]

Document the hold. Let's make sure we are also documenting the how to track part.