This is just my observation from running the patched node for the past month which included a few hours(?) the mempool was actually clearing 1 s/vb txs. I do recommended running a patched node in parallel if you can to see for yourself. I just added 2 lines to lncm/bitcoind Dockerfile to get it going.