if you're attempting a reorg you aren't mining on the latest block, you are mining on a previous block to attempt to remove the latest block.
I think I see what you mean now, by "mine the next block" I think you mean "build on top of whichever block was most recently mined prior to the reorg," whereas when I say "mine the next block" I mean "increase the number of blocks in the blockchain, regardless of the contents of your blocks."
Assuming I understand you properly now, I think that if a set of miners tries to steal from a drivechain, and another set tries to stop them, that is an example of contention. Bitcoin's censorship resistance relies on its ability to work amidst contention. If it can't operate in a situation where one group tries to censor another group's blocks, then it is not fit for purpose, because that is part of the real world bitcoin is designed for.
What I think should happen in that case is this: the set of miners with 51% hashrate will win by slowly extending the chain farther than the other set of miners. The other set of miners will have to capitulate because at some point the cost of performing their attack exceeds the reward. The moment that happens, it proves that bitcoin's method of working through reorgs works. If it doesn't happen, and we just get constant reorgs over and over, that proves that bitcoin's method of working through reorgs has a problem that must be fixed.
Either way, I don't think drivechain creates this problem, rather it helps us prepare for it ahead of time, because if the real world's economies adopt bitcoin, such contentions are inevitable, and I for one would like to know that bitcoin can handle them.
this is just generally incorrect.
PoW is not a method to achieve consensus under a different set of conflicting incentives. It is not a "method of working through reorgs".
It explicitly sets out to make it more profitable to mine on top of the next block according to a discrete incentive model. If reorgs occur then this incentive model is explicitly broken and the system is proven unstable ie. useless.
It's a careful balancing act and drivechains challenge this. "Preparing for this" is preparing for armageddon. You can be prepared all you want in the end the outcome is the same. We all die.