It's unknown to what extent the directional use will manifest as a problem. A few mitigation ideas exist already, such as negative fees to encourage use of routes in the direction of rebalancing the channel, or rebalancing by means of other payments as needed.

I don't think that receiving one's paycheck to a Lightning channel is a valid shortterm vision, nor is full reliance on Lightning Network for all payment purposes a palpable near future use case.

Let's be honest here. LN is something totally new, and a few courageous adventurers will start using it on a live network with minuscule amounts of money. As we observe how that goes and what assumptions turn out to be true, we'll be better able to judge the use cases it satisfies and onboard more users and value. Then we can further observe how that improves reliability and functionality. We neither have to switch over fully, nor in a short timeframe.