Blocksize is an extremely overrated issue.
Doubling the blocksize at most will mean maximum theoretical transactions might increase from 7 to 14 transactions per second.
It won't make a difference. It won't solve issues with fees or unconfirmed transactions.
Its not worth it in terms of risk versus reward.
Of course it will make a difference. Fees and the number of unconfirmed transactions correlate quite well with block space usage. But a conservative one-time increase of say doubling to 2MB is not enough. Segwit is not enough either.
We need a something like the 3rd option, but don't think Core would endorse anything like that on-chain. Choosing the algorithm for that shouldn't be done by the developers, they aren't the proper stake holders here. Developers shouldn't have any kind of decision power except by way of technical consulting.