TLDR - the issue is with mobile Vs. Desktop reporting; desktop reporting is better than mobile performance, but the overall mobile grade is better than Desktop.

Why the issue is not a bug - the Web> Mobile receiving a better score would be expected behavior because grades are based on a percentile range for a business's industry. The report doesn't look at relative performance between desktop and mobile but rather at their industry category.

From the customer's perspective: You would agree that the graph and the score for a human being don't match. Therefore a potential customer will be confused or believes the data is not reliable. It might be mathematically correct based on the theory the score is being calculated. However, this might please a developer but can hardly be understood by a prospect customer.