Hmm, not a fan of the 'fix' at all. Sadly, it was designed with an exploit in mind and how to prevent it, not with how the power *should* work.
Some realistic examples of a paladin trying to play according to character, not trying to exploit the rules, let's see how this fix applies:
1) Tough battle, Paladin and Wizard are the only ones left, vs two brutes. He's fighting one brute and those two are slugging it out. The second brute moves next to the injured wizard and is quite likely to kill him in a round or two. The Paladin issues a "Divine Challenge" against the second brute - "Come here you villain, fight *me*!". It's not supposed to be like a fighter's marking, this one is based on divine power, and the courage of the paladin. Calling out a second foe to put his life in danger to save another is about as classic as it gets. The brute has to decide whether to take his lumps and try to dispatch the wizard quickly, or engage the paladin. The paladin can't simultaneously be next to both foes. With the new 'fix', this use of the challenge just got broken.
2) Same situation, but foes are across a chasm and it's a ranged battle. Paladin is weak at ranged combat but whips out a crossbow and does his best. Same divine challenge issued, "Shoot *me*!" Now because he is attacking, the challenge works. Technically. What really happens is that of course the archers that did not gets challenged shoots at the wizard. No improvement harm over unfixed challenge.
3) Paladin is trying to be a forward-deployed bottleneck. One creature happens to slip through and if headed toward the squishy warlock in the rear. "No, I said fight me!" But he can't engage and is busy attacking others. So the fix reduces the effectiveness of the paladin as bottleneck.
4) Speedy on-the-go ranged attacker loves to move faster than the party can and take a shot at the wizard. Original divine challenge, paladin says "oh no you don't, fire at me!" Now... because the paladin is incapable of getting physically next to the foe and has no ranged attack, his divine challenge will never work in this situation.
In the absence of a "rule" I seriously doubt I would have any problem either as a DM or a player in seeing where the divine challenge would or would not apply. Any player gaming the system and ignoring where his divine power is coming from would learn very quickly that nonsense won't work.
The critical point here doesn't seem to be whether the paladin is attacking or engaging or not - he's trying to draw attacks, not mark the same way the fighter is. Isn't the essence here whether or not the challenged foe can get to the paladin and/or attack him without being 'stupid'? It's more like a charm spell in that sense - swing/shoot at A instead of B when it doesn't harm the target - no problem. Take a step towards me even though the fighter will mow you down, or take OAs because I'm challenging from the back row. A "duel" where you take ten paces then fire isn't a fair duel if the person is standing in a mine field or if the opponent is hiding behind a wall.
I don't know if the following would work, but seems more in line with how I picture the challenge. It works for the situations above and the runaway paladin exploit.
The divine challenge will be voided at any time where the challenged creature is unable to attack or move to attack the paladin without putting itself under attack.
(What this would shut down would be a ranged paladin challenging a foe pinned down by a fighter, which I think is against the spirit of a the challenge anyway. An alternative would be voiding the divine challenge if as a result of trying to attack or engage the paladin the marked creature is attacked by an ally of the paladin)
Conclusion:
* Requiring the paladin to attack or engage the challenged foe absolutely eliminates his ability to draw a non-adjacent foe to him to protect another party member when he's engaged with another for or otherwise stuck in place. That seems totally against the differentiation between a divine challenge and the martial one.
* Attempting to add an "elegant fix" to an exploit as a rule instead of using common sense is doomed to fail if you don't consider "normal" situations affected by the rule change and not just players trying to game the system.
* It's bad enough to have to house rule a vague rule to fix an exploit, but even worse to house rule a specific rule to allow the character to act normally.