I am publishing this so that the relevant code and chronology are available for others to examine.
On September 16, I received an email from someone at Ente thanking me for FairScan and telling me that they had added a "scan" feature to their Locker app: the original version of a blog post credited FairScan for its segmentation model and explained that they built the rest "from scratch".
I was curious and looked at Ente's code: I found several image-processing implementations that appeared to closely follow corresponding FairScan implementations.
