They could have done such experimentation in an unreleased fork of the browser. Or if they needed wider testing, they could have released this separate, experimental branch and clearly labeled it so that people would know what they were getting into if they installed it. Mozilla didn't need to force Pocket on so many Firefox users who wanted absolutely nothing to do with it.

Why the rush to distribute it first and figure it out later? Couldn't they figure out the API extensions required during development, but before shipping?

Pocket? What exactly is the integration doing that the addon wasn't able to do?

This is a real, non-rhetorical question: I started occasionally using the addon when pocket was still read-it-later and never changed usage patterns since.