I have no inside info on this incident and haven't worked for Google in over 4 years, but the description of the 14-day 2019-only issue sounds like it would have affected only a subset of users whose data they included in a sample for troubleshooting purposes. The sample rate they use is probably far below 100%.

(Once again: this is speculation based on general firsthand knowledge about how Google operates, but I know nothing specific to this incident.)