I may flirt with personal opinions here but as always alternative answers are welcome if people disagree with how I classify things.
Anytime there is an attempt at a consensus rule change and there is resistance or opposition to that consensus rule change Bitcoin is then in chain split territory or as close as Bitcoin gets to a "war" over the consensus rules like we saw with the block size wars. Hence I and others have argued that consensus rule changes should be made (to the extent that they are made at all) infrequently and with a very high bar of testing, review and community consensus.
Let's say that an attacker wants to deploy a softfork, that isn't desired by the majority of the Bitcoin users. Can he enforce that softfork in a UASF-way, essentially flooding the network with nodes that support it?
The scenario you describe of a lone attacker with lots of sybil nodes isn't particularly concerning. At best it could cause disruption on the P2P network but it isn't going to succeed in changing the consensus rules on the blockchain and on the Bitcoin network without community adoption. Andreas Antonopoulos has described Bitcoin as having "constituents of consensus". We could argue over how to break these constituents down but they clearly include (in no particular order) economic full nodes (nodes verifying their economic activity), protocol developers, miners, exchanges, wallets and protocols built on top of the base layer. If you are attempting to change the consensus rules and you don't have the majority of these constituents onboard then Antonopoulos has described it as attempting an economic blockade in the middle of the ocean. Or in other words you may shout but no one listens to you and just sails around you. You may fork yourself off the Bitcoin blockchain and the Bitcoin network but you won't succeed in changing the consensus rules for everyone else.
Hopefully we won't have another consensus rule "war" anytime soon (ideally ever). During "war" like scenarios there are "fog of war" like dynamics and misinformation deliberately spread to cause confusion. What constitutes a "war" is also unclear. A single individual in the middle of the ocean declaring "war" is ignored. If it starts to look like the constituents of consensus are split and one side is going to attempt to force through their consensus rule change against the other constituents' wishes then we are definitely in consensus rule "war" territory again.
Personally (opinion alert) I think there will be future soft fork(s) attempted with the vast majority (ideally all) of the "constituents of consensus" onboard and comfortable that the change being made is optimal and worth taking the chain split risk for. If we ever do attempt it it will be all our responsibilities to ensure accurate information is circulated and any troublemakers from say altcoin communities don't succeed in turning that consensus rule change into a "fog of war" like scenario. But as I said I think there is general acceptance at this point that the technical and community consensus bar should be very high before embarking on an attempted consensus rule change and that, perhaps more than anything else, protects against chain splits occurring out of nowhere.