Also worth noting though that "extra" energy per-tx increases the security of that tx. You aren't just paying for the system to function, you're paying for the system to be resistant to bad actors.

Only so far as the majority of the miners are not coordinated.

With the current concentration of mining capacity in China, that's not exactly a sure thing any more.

The point is that the energy they are spending on mining disincentivizes them from doing anything except using it to earn Bitcoins legitimately. If they waste that energy attacking the network instead, it would be less profitable.

Yes, but the limit is chosen to keep a balance between scale and decentralization (keeping the cost of running nodes low).

It can be safely increased over time as bandwidth, storage and processing costs go down.

For now there's still a lot of viable efficiency enhancements to be implemented/adopted so increasing the limit isn't strictly necessary.