It's plenty enough, and it's quite rare to have a good reason to change it. The main purpose of the mempool is simply to help transactions be propagated for the purpose of being mined, by storing what transactions have already been propagated. Transactions with low fees aren't going to be mined soon, if ever, so there aren't good reasons to propagate them or store them in mempools. Propagating them is generally just a waste of your peers' bandwidth.

Huh? The defaults for both those options work fine for the vast majority of people. If anything, we're better off if a large number of people run a lot of smaller nodes. That makes deanonymizing Bitcoin transactions by tracing them to their origins much harder than a small number of big nodes.

Having nodes that aren't publicly accessible improves privacy for everyone by making it harder to identify peering connections, again making it harder to trace transactions back to their source.