I've seen conversations on this since at least 2005. There were even proposed patches every once in a while, but never any consensus. Anyone care to comment?
Well, as you said, there has never been any consensus.
There are basically two pieces to the puzzle:
1. What metric do you use to represent correlation between columns?
2. How do use collect that statistic?
The option that always made the most sense to me was having folks ask postgres to collect the statistic by running some command that marks two columns as correlated. This could at least be a starting point.