It's not clear what the actual issue was. My best guess is that Workers KV was in transition to B2 from a combination of GCS and "a storage provider" before. A step in transitioning to B2 involved removing the "a storage provider" and relying just on GCS for some period of time because of consistency concerns. Today's GCS outage was unfortunately timed and they didn't have coverage to support GCS being down.

Separately, their dependence on their own products results in a SPOF where now, theoretically, if B2 goes down so does over half of Cloudflare's product suite. Ideally those outages can be limited to a single region at a time but that's still a massive blast radius from a single service. I completely understand them not wanting to invest in competing products as a form of redundancy though.