I know MT is supposed to be standard but if you have one company making MT stuff using metric machines and one company using inch machines is there not a chance the MT could differ enough between the two that it could be measured? That would also include the competency of the programmer and to what decimal place they programmed the machine cutting the MT.
MT (Morse Taper) number is basically an angle when you get down to it. It came about for historical reasons & has perpetuated to current machines. Whether expressed as inch/foot, inch/inch, mm/meter, decimal degrees, Angular Deg-Min-Sec.... its the same thing & should be metric/imperial agnostic. What you might be asking is how accurately is it made which is a tolerance issue & machine/operator certainly comes into play. If there is no stated tolerance then you have to cross your fingers. If its a stated tolerance & hobbyists class & from a far away land with certain reputations & costs $50, you might also have to ask if you can believe it. And if you even have the means to even validate it. Generally speaking, centerless grinding if done properly, should be superior to anything cut off a lathe in both accuracy & finish, unless the lathe part has been lapped.
The Ebay test bar I bought was made in India which I figured was probably throwing money out the window big time. The USA made ones were way out of the budget & these are generally not high volume items. I cant recall exactly now but I think mine said 'centerless ground within 0.0002", but no stated tolerance as to the MT angle. So now we get into relative tradeoffs of unknowns & I suspect others may have a different opinion. I felt reasonably confident that my MT spindle socket was ground sufficiently accurate and ground simultaneously as the other spindle features that dictate chuck fit-up & concentricity. My DTI said it was concentric at increments down the bore (not egg shaped). Several decent quality MT tools fit very well as evidenced by blue-ing. When the test bar arrived (in a hopelessly inadequate bubble wrap envelope but semi-padded box) I checked the socket fit again by bluing & it was exceptional. Diameter readings down the length of bar confirmed the tolerance to the best of my devices (10-ths reading micrometer). I don't have a granite surface plate / datum surface to detect bar bowing but I semi-faked it by traversing down the bar with 10-ths indicator, rotate 90-deg & repeat. Again it checked out for my intended purposes.
With these basic quality checks /- confirmed, what is important to say is a test bar in this application is a glorified extension stick. We are using simple trigonometry to our advantage by comparing bar end position over an extended length: 12,18, 24" away (depending on which one you buy). The longer the length, the more it exaggerates axis displacement. Its as simple as that. So even if the bar diameter varies by 0.0002", its not nearly as significant as if bar is pointing in/out/up/down by say 0.005" over its length. Now we have a starting reference point. Assessing this same geometry discrepancy by cutting a coupon is do-able not quite straightforward because cantilever gripping a 24" long piece of whatever metal & cutting a slice off, we are left with a question: how much of the noted diameter difference (taper) is due to geometry issues and how much is due to cutting force variations, material deflection etc. Is 0.005" difference comprised by 0.003" geometry + 0.002" cutting forces? Or maybe its 0.004" cutting forces + 0.001" geometry. Lots of opinions, significantly less real world examples & hard data at least that I have seen. if someone has something to share, lets all have a look. The test bar in the socket provides some convenience in that you have the means to make the HS adjustments right then & there with DTI residing on the bar.
I'm not saying cutting has no place, it certainly does because it reflects real world conditions where the lathe is doing what its intending to do. But IMO, if its a bolt on HS I see value in validating what might be overriding dominating problems with a test bar first & followup/validation cutting second. In the end we all have to try & sort out amongst the chatter what makes sense & find our own way.