This demonstrates:
1. Lightning isn't perfect as per the 2015 original specification
2. It can be improved, and the developers know what the weaknesses are and are thinking carefully how to redesign optimally
It will take time, but the end product should be easier to use than regular Bitcoin clients. And this will be huge for Bitcoin's competitiveness as a currency, it could be like using gold with a debit card, where the actual bullion is under the complete control of the individual.
I think we all know that this person did not pull enough liquidity from the system to make it technically impossible to route payments that were not marginally large. It's just that there wasn't sufficient liquidity to overcome the shortcomings of imperfect routing. I'm skeptical that full client side routing may end up not being practical. Perhaps a hybrid system where a satisfactory level of privacy is achieved through onion routing while other hops in the routing are a collaborative effort? Not sure, just thinking out loud.