If you cannot establish high precision on your bench equipment and know that you have, you will not have precise results, and you are only going to chase your tail. The horizon method is precise in function, as it uses an established infinity point, not an artificial one, and you can see when adjustment is out. You do not have to use the horizon; it can be a tree on a hilltop several miles away, for example. I prefer to use the moon at night, as it is a high contrast true infinity target. Using the focusing screen image to set mirror angle might seem crude, but it's not. It's a simple and real-world method that works very well, because it relies on a non-variable target that requires no calibration. Once precise mount-to-film plane distance is set, a ground glass at the film plane establishes the point of precise focus, then the mirror angle and screen height are matched to it.
Not to denigrate the bench method, which is much more practical in a commercial situation. But to build your own precision optical test setup requires a precise means of fabrication. How are you going to precisely determine center of the lens mount? How are you going to know that the laser beam is precisely perpendicular to the film plane? So using a real-world infinity target and calibrating by observing focus on the screen is faster, easier, more practical, and more trustworthy.

Some points:

> Achieving some percentage of factory precision really isn't good enough. It's either in tolerance or it's out. I wouldn't want any repair person deciding that out of tolerance by 8% or 5% or even 2% is good enough. Don't worry about setting up an optical bench when you can use other means to set mirror angle- instead get some better precision tools, like a depth micrometer that reads to .01mm.

> You should bring that .04mm down to .02 or better, to give some room for error. Using the caliper is iffy, and the closer it reads to exact, the more likely you will be in tolerance. If your present measurement is precise, then you are out of tolerance, because the factory tolerance is .03mm. It makes no sense to start out not knowing where you are, because you will never know where you will end up.

> Make sure the focusing screen is flat, as stated in the manual. This is essential.

> With the screen height screws all the way down, the shortened optical distance from the lens mount flange means that the screen image should be out of focus 'beyond' infinity. By turning the lens focusing ring off its infinity stop you will increase that distance until infinity focus is reached somewhere on the screen, and you can adjust the mirror angle until all parts of the screen image are in focus. That's your final mirror angle setting.

> Turn all four screen height screws the exact same amount to raise the screen to its true setting for infinity, using the ground glass at the film plane to establish true infinity focus and adjusting the focusing screen to that.

> If your lens mount to film plane distance is off, properly adjusted lenses will not focus to exact infinity at their infinity stop and any lenses you adjust to the body you are working on will not be right on other bodies. If you have access to a properly adjusted lens, you can use it to check the lens mount to film plane distance. Use it with it set to its infinity stop in conjunction with the ground glass at the film plane to check infinity focus-if it's not right the body needs adjustment. You can also use a known good body to calibrate your lens' infinity stop and then use that lens to check the body you're working on.