1. That doesn't mean killing it is worse, and I'd go as far as to say this case is objectively better to let rot. Imagine the outrage and brand stink that would create for Google, who is already known for killing things liberally, in a product space that is focused on longterm reliability. By letting it rot, people who deployed and are just leaving on autopilot don't have to do anything, and GCP offers lots of alternatives that can use Python 3.

2. Bad product management / development speed seems to be the case here, not rot or killing, as it seems it's actually not stuck on Python 2 anymore, or for long [1]

3. Even if neither of the two things above were true, I have seen no bad press on GCP due to this, and I'm a heavy GCP user, though just not in this niche. I've used Python cloud functions as well as GAE though.

[1] https://cloud.google.com/appengine/docs/standard/python3