I think it is just an honest mistake. Some sort of glitch. I don’t think anyone can turn that into a scandal.
How likely is it that two polls, 3 weeks apart, have the exact same number of likely voters in the sample (1,347) and shows the exact same numbers in two races (R 49 - 0 45 / Smith 48 - Casey 46)? That exact match of likely voters is really the giveaway. The chance of all the numbers, including the number of likely voters that made it through the screen, to be identical, would have to be astronomical.