Note that there is actually "only" around 65 vMB of unconfirmed transaction in the mempool. The 360 MB figure you are quoting is the dynamic memory usage (i.e. the RAM usage) of these transactions after they have been deserialized. If no more transactions were added to the mempool, and assuming an average block size of ~2 vMB, then it would take 33 blocks to clear this backlog, and not 180 blocks.
Mempool has 80vMb with higher purging rates than others, make that ~120vMB at 500MB with purging everything under ~1.2sat/b, so one can easily assume that if a bit of it is cleared a lot of the 1sat/b sanctions that have been purged will come back, it's highly unlikely that everyone that had a broadcasted 1sa/b tx just went and upped the fee. Then add the extra 37 blocks we had due to more hashing power and it's a lot worse than just 33 blocks.
3 days till the weekend, although the mess started also during the weekend!
Does anyone have a link, graph or something that counts these "minting' transactions/day?