No, it says “not necessarily per product unit”. Your characterization of the abstract is incomplete as it doesn’t definitely state what you’re claiming it states. It’s also a euro meta analysis, not a US analysis, so extrapolating your oversimplified conclusion is even more of a stretch since we’re talking about the USDA. I’m more concerned about carbon, water use, pollinator collapse, and a host of other metrics than NOx (which is a function of diesel emissions standards and crop yield, and can be fixed independently).
How can you possibly think the US military, or any sovereign country, will magically spend an extra $165B a year on meat a year if all of the current consumers magically go vegetarian? Who exactly is going to eat a bunch of extra meat? There will just be fewer meat sales, period, ignoring a short term price drop if everyone magically goes vegetarian on the same day.