Keep in mind that load sharing is not speed matching. It is very likely that two locomotives will run at different speeds when uncoupled. Think about it....the principle of load sharing is based on the locomotive modules detecting a load, and then distributing it across the consist. If a lead locomotive isn't coupled to anything, then it detects no load, and therefore tells the following locomotives to contribute less to the overall pull. So yes, the following locomotive will usually run much slower than the lead if they are uncoupled. How much slower depends on the individual locos and their settings.
You'll notice that the effects of load sharing are much easier to see when pulling a longer/heavier train. Try it and you'll see.
That being said, there have been a couple of locomotives I've built and installed Railpro in that I had to dial in the MFLC manually in order for them to play well. These were frankenstein locomotives though....motors were not original and much work was done to the drivetrains.
Also, going along with Alan's comments, I will typically run my MFLC settings three times on a new locomotive to get the most accurate one. Does it really matter? Probably not, but I do it anyways. If you think about it, the MFLC will drop slightly after first time due to the wheels/rails getting mildly polished from the spinning. Running the locomotive for a bit beforehand to warm up the motor and loosen the gears a bit may also make a difference. It will likely be a small difference, but a difference nonetheless.
Out of curiosity I have re-run the MFLC on locomotives that had it set years ago, and remarkably the values have largely been the same, maybe within 10 mAh of the original or right on.
-Kevin