Moral obligations ought not to depend on an agent’s character. We ought to insist that an agent with a bad character act entirely like an agent with a good character...
That seems inadvisable. Cases like the bad squash loser show that we ignore our moral failings at our peril. For a more high-stakes example, suppose God offers me a deal that's a freebie for a saint, but insanely risky for the rest of us:
(SaintOrSinner) God will save one innocent life now, but if I ever fail to meet a moral obligation for the rest of my life then a million innocents will be tortured and killed.
Should I take the deal? If I'm a saint (and know it), then it's clearly obligatory: I can save a life at no real cost, since - as a saint - I'm sure to meet my later obligations whatever they may be. But since I'm not a saint, it's very clearly impermissible -- I'd almost certainly be condemning a million innocent people to torture and premature death.
More generally, it's fallacious to move from the fact that it would be good to do P and Q to it would be good to do P. Remember that evaluations of act-aggregations and of individual acts may diverge:
If my future self cannot necessarily be trusted to do the ideal thing, this could radically alter what current decision would be for the best. Suppose my currently φ-ing could lead to (i) the best possible outcome if I were to follow this up with a series of acts S which I could, but actually won't, perform; and (ii) the worst possible outcome otherwise. In those circumstances, it is not best for me to φ -- doing so would have very bad consequences, due to my subsequent failure to S -- even though it's part of the best possible life. The divergence arises because at any given time I can only choose how to act then; I cannot perform a lifetime's aggregation of acts with a single decision.
So we need to be clear about what normative question is being asked. It may be advisable both (i) to do P-and-then-Q, and yet (ii) not to do P (because one is unlikely to follow up with Q, and the consequences of P without Q would be disastrous). This is perfectly consistent, because the two answers address different questions: one concerning the aggregation of acts that would be best, and the other concerning what present individual act would be best. It's just a basic fact that in non-ideal agents, these may diverge.
1. Although we will typically be more interested in deliberating over individual acts than collections thereof (since we can really only perform the former), there may be exceptions. See, e.g., my complaints about illegitimate appeals to 'Political Reality': they may be fine from an individual perspective, but sometimes we want to deliberate from an explicitly collective perspective, about the question of what 'we' (rather than just 'I now') should do. This shift in focus can be vital for breaking out of a bad equilibrium.
2. My old post on 'Accommodating Unreason' offers some reasons why, at least in low-stakes contexts, it may be better to treat people as if they are morally more reliable than they really are.