I think FlexTrans is a good solution. As a hard fork it doesn't include the immense technical debt that comes with SegWit. A malleability fix will enable bells and whistles like atomic swaps and LN (the way it has been developed so far - I know it is possible to implement without a malleability fix), which would be a net positive.

I'm personally not opposed to L2 scaling solutions. It's the crippling of on-chain scaling to push L2 solutions that I take issue with. Let LN compete with on-chain transactions directly, without an arbitrary restriction limiting the network's capacity.