Well, I'm a big subscriber to the "if it ain't broke" approach. No way would I mess with the muzzle of a firearm I would describe as "extremely accurate"
Just think about it in terms of risk:reward. How much more accurate do you believe the firearm could be? That's your potential reward. How much less accurate do you think it could end up. That's your risk.
JohnKSa, I haven't read that column, so I don't know what the methodology or sample sizes were, and have no way of telling whether the numbers are meaningful or not, the examples in your quote show increases in dispersion on the order of 60-70% of the premutilation figures. Pretty big difference.
Oh, and cleaning up the crown should only take off what's needed to remove the visible damage. How much you lose is directly related to how deep the nicks are. Given your description, I'd guess a few thou is all you'd need to clean it up. Not enough to visibly shorten the barrel.