What you're saying there is fair enough mate, but I think LT's point is that IF we're going to remove "minnows" (and whether we should or not is another question) then we should remove all of them, not just the ones that suit our argument.
But his point is myopic. Success against minnows does not matter as much (notice I didn't say "doesn't matter at all"), because the inherent argument is that they weren't good enough to challenge him. But failure against them - over 12 tests - is very relevant. It has nothing to do with suiting an argument IMO, it has to do with getting to the truth.
It would seem intellectually dishonest to argue that failure against minnows doesn't matter but success should.
In fact, if one were to remove all 3 it would still hurt Sobers' career figures. If that was all the intention one had then it's fine. But there is no reason to disregard his failures against minnows.
On the whole, his record is still up with the best there has ever been IMO. Personally, I think every batsman has that kind of record against one opponent. Every batsman has played some minnow too and depending on how many and how often we should acknowledge it. I don't think it's a very strong argument to count England, Australia and New Zealand but disregard the others, though. He was still the best batsman of his era and everybody else played them too. It's only when you compare cross-era that it gets iffy.
Sobers averaged 57 overall, but if you do compare him with others I think it's fair to realize that it probably does flatter him a bit. I mean, imagine if Lara had a career, and the sides he played against were: Australia, S.Africa, Zimbabwe A, Zimbabwe B, Zimbabwe C. Even if you think they were better than Zimbabwe, IMO they weren't really significantly different and it shows that a greater proportion of matches were played against much weaker opponents.