They have the incentive but they are not configured accordingly. Not all of them. That's why dust/spam/no-fee and "stress test" txs get processed.
Have you used the network lately? No-fee transactions aren't getting included, and definitely won't be once priority is removed. Dust and spam should not be subjectively determined, fee per kb is all that matters. Miners have a very real incentive to keep their blocks to a reasonable size, it's called orphan risk.
The 1MB limit is definitely helping to prevent abuse. Otherwise the attack vector and the willing miners are there for exploitation in a 5MB, 10MB, 20MB block scenario.
Straw man, the miners have said they are ready for 2MiB, maybe 3, so that should be the starting point for our can kick. See jgarzik's new proposal out today, it seems a very reasonable compromise between the harder line factions of the two camps.
I do not believe that BTC should stay at 1MB. If it was up to me, it would grow depending the actual organic growth of the network* and other considerations that have to do with security, centralization etc.
* by that I do not mean the number of transactions. This is a pretty useless metric if one can start issuing tens of thousands of txs.
Thanks for clarifying that. If you have suggestions of how to measure the organic growth outside of transaction levels and address usage, I'm sure the dev's would love to hear about it.