We can argue semantics all day, but the bottom line is that the stat does absolutely measure replacement level over same position players. That's exactly what the output measures... the number of wins above a replacement player at the same position. It's astounding to me that anyone is disagreeing with that simple statement. That's the whole calculation in a nutshell. We can argue semantics at the 'purpose,' but that is in fact what the stat does.
This is basic economics. Some products might not have as good of quality as another product. But supply and demand might make that product more valuable despite being of a lesser quality. WAR doesn't aim to measure the quality of the product, but rather add in the supply/demand value aspect of the positions.
One last time... Ryan Hanigan isn't a better baseball player because he plays at a position with lesser talent. All that means is he's a more valuable baseball player. WAR is meant to measure value, not necessarily talent/production. You don't need to know replacement level to compare players across different positions if all you want to know is who the better baseball player is.