LN would be cool, but we don't even need it anymore. I can send a BTC transaction with a 5 cent fee and an ETH transaction with a 5 cent fee, and the BTC transaction would actually confirm much faster, in most cases (and in current mempool conditions).
That’s not relevant. I think big: I want for Bitcoin to compete with Visa. For that, we need to increase peak TPS throughput by five orders of magnitude. The blockchain will never do that—not unless Bitcoin Jesus blesses us with gigabyte blocks /s (got a supercomputer for your node? /s). I expect that Lightning will, after it matures.
For an engineering analogy of why Lightning is important, see this article (forum discussion where I found the link).
I think that BTC dominance will always remain a given - for various other reasons.
Certainly, Bitcoin will remain dominant. But it could not do so without the best developers, the best new ideas—and scaling both vertically (Lightning) and horizontally (sidechains).