If the profitability of mining BTC is better than that of B2X, then those 15% will turn into 90% within a day.

I believe it will take quite a bit more than a day; the situation needs to be stabilized before it's clear what is most profitable. There is a delay from the block is mined until the coinbase can be moved. Exchanges most probably will disable all deposits and withdrawals for some period surrounding the fork, there is no profit to be had unless the miners can transfer bitcoins to the exchanges. Market prices for the two chains are likely to move a lot in the period surrounding the split. I do expect the miners to stick to the agreement for several days, maybe as much as two weeks ... but of course, in the end the market talks.

I'm surprised that the Bitfinex future market gives the 1MB-branch such a solid support. However, it's wrong to assume the post-fork token price on the two chains will mirror the bitfinex futures; the price also reflects other risks, like the fork not happening at all (if I've understood it correctly, the BT2-tokens will become worthless if Core folds and merges the 2MB pull request - even if that means a full victory for S2X). The Terms and Conditions also specifies that the Terms and Conditions can be changed at any time by Bitfinex, something that is quite unacceptable in my view.

Worst case for BTC is it might have 2 weeks of slow transactions.

No.

Bitcoin is designed so that the average time between two difficulty adjustments is two weeks - in reality it has usually been less than two weeks. However, those "two weeks" are not measured in clock ticks but in the number of blocks mined. This means that if going from 100% to 20% hash rate, those two weeks will turn into ten weeks. With 10% hash rate, those two weeks will turn into twenty weeks.