What will happen once all on-chain scaling methods are already on place (so let's say, segwit at 90% usage, schorr sigs, MAST...) and the fees start getting too high even after deploying all of that? How high would the fee need to go to consider a blocksize increase?
Define "too high". Clearly, everyone has different definitions of how much is too much, people are paying historically high average fees at the moment which are clearly not too high for those people.
As for on-chain size increases, those need to be preceded by study into off-setting the additional processing burden higher blockweight limits represent. Segwit is somewhat of an example; allowing signature hashing to scale linearly mitigated (but did not offset) the increase to a 4MB block limit.
Aggregated Schnorr signatures will positively off-set the effect of larger blocks; not only will they permit more space within the present limit, but the much reduced processing required for aggregated signatures means larger block sizes should be safer (but only for block-space containing aggregated signatures, much like how >1MB of block space is only available to Segwit transactions).
And under what scenario isn't a blocksize increase a total clusterfuck which would end up with 2 coins?
Segwit. Increased block limit to 4MB. No clusterfuck (or continuation of a non-Segwit hardfork) occurred.