Segwit eliminates the size limit and replaces it with a weight limit. The weight is computed so non-witness data (e.g. the above outputs) count 4x as much towards the limit as witness data (signatures). The fixes the incentives by balancing the costs of signatures vs outputs to be roughly equal.
It 'fixes' the incentives only in cases where nodes throw away data needed for any future validations of the transactions. Whether or not this 'pruning' is a good idea is a matter of reasonable debate.
It's also how segwit achieves a capacity increase in a way which is fully backwards compatible with old nodes.
Turning fully-validating nodes into non-validating nodes is a rather funny definition of 'backwards compatible'.
Finding a way to address the UTXO incentives issue was a major sticking point and breakthrough that made it possible to get many people to support any capacity increase at all.
UTXO set size is some function of (# users) * (# of addresses holding value per user). As far as privacy is concerned, best practice dictates distributing your value across several addresses. Are we to follow The SegWit Omnibus Changeset with a recommendation for each user to hold all their Bitcoin on a single address? Privacy be damned?
As to why Bitmain would complain about it, I am aware of no sensible reason.
Introduction of a new fixed centrally-planned variable? Preferential incentive for offchain transactions over onchain transactions? Myopic much?