At first I was convinced by your argument, but then I tought to apply it also to the voters.
Both parties would be probably better with a lower turnout (less undecided voters to convince, less effort required to campaign); but it does not seem correct to conclude that we should not encourage voting. An high level of engagement of the local popolation in politics is supposed to be a good thing for democracies, even if it requires more work.
I do not think that donations to political parties should be tax deductable. But I hesitate on the abstract principle “Do not encourage prisoners dilemmas”. Sometimes it is good that everyone makes more effort, because that effort produces something valuable. Two rival pencil companies “play” in opposition to each other (in the estreme case in which the pencil demand is totally inelastic, it would be a zero sum game, because if more people buy pencils from company A, then less people buy pencils from company B). But this does not mean that the state’s regulations should encourage monopolies: monopolies are (generally) not good.
Depending on your utility function, there are situations in which you should wish to encourage a prisoner dilemma between two factions, if this yields benefits to you or other people.
Maybe your principle applies when the relevant model really has no third party which gains from A and B fighting. But I can not think to many such situations in the society (even in wars, there is someone (weapon sellers?) who profits if the belligerants fight harder).