Here's the elephant in the room, about why it's so ****ing hard to rate all-rounders.
The truth apparent to anyone who watches Test cricket is that (excluding Bradman for obvious reasons), the very best bowlers just blow out the very best batsmen in the world for value. McGrath vs Lara, absolutely laughable comparison. For the top 6-7 bowlers of all-time, they could make an impact towards match result in their team's favor with a consistency that batsmen could never dream of. Even something like the top 15-20 bowlers all time, are going to be more valuable, in my estimation than the very best batsmen, although a slight step down from the top tier. After that some bowlers' value might be more comparable to batsmen, but in these tier it's leaps and bounds higher.
Batsmen are inherently inconsistent, and at the very top they have a very granular difference from one to the next, in their ability to effect matches, unlike bowlers whose top end impact seems to be almost limitless.
So you combine those two attributes, and it's hard to really rank "value" of an allrounder as compared to "balance" of an all-rounder. No one other than Imran could possibly compete with Hadlee on "value", but he doesn't have the "balance" between the two attributes which we see in all-rounders.
Looking just at batting and bowling, that's why I had Sobers "valued" as the 7th best all-rounder in my initial list. But that isn't how we really rank all-rounders. The real way is something like a "rating" of 1-100 for batting, 1-100 for bowling, then add them up. By that listing I'd then have Imran and Miller at the top, followed by Sobers, at a more reasonable 3rd (he's in the same tier as the above if you factor in fielding too).