If theymos were to follow the same rules regarding allowing the discussion of XT/classic-like proposals in the Bitcoin section as he does these kinds of features, then discussion about 0.12 should belong in the altcoin section and 0.12 should not be advertised/announced by the forum.
If RBF was a hardfork, there probably wouldn't be sufficient consensus, and 0.12.0 would not be Bitcoin. But RBF and the new mempool policy aren't changes to the consensus rules at all, not even a softfork. The utilitarian reason for the hardfork policy is not because I think that somewhat-controversial changes are inappropriate in general, but because in the specific case of contentious hardforks (and soft-hardforks):
- Bitcoin is split into two pieces, which is extremely damaging to Bitcoin.
- There is an attempt to coerce people who agreed to a certain set of "immutable" core consensus rules into abandoning these rules.
These two problems don't exist for softforks or network policy changes. Old nodes continue to follow the rules they agreed to, but they will still be on the same network/currency as new nodes. Likewise, for non-contentious hardforks, Bitcoin is not split, and very few people are disenfranchised.
They're also very good features. The mempool policy is especially useful. I had one node on a somewhat low-memory system that would often crash due to an overly-large mempool, and 0.12.0 fixes that. And RBF will allow for solving the "stuck transaction" problem in future releases. (If I thought that some changes were especially bad/damaging, I'd probably point to some consensus-compatible fork of Core like LJR instead.)