I believe the "won't DoW when already at war, except Monty" isn't a hard rule. It's just that being DoWed/right before DoWing someone the AI will (re-)evaluate their situation, and only Monty is insane enough to start plotting a war even when he's already in a war. Any other AI, assuming a relatively sane situation, won't consider throwing themselves into a two-front war. Catherine, in this case, must have been in such an overwhelmingly superior position (at least in her mind) that she went through and DoWed you anyway. AIs can definitely be opportunistic and snipe cities, it's just that they very rarely choose to do so while already fighting a war.
Incidentally, I should maybe add that Catherine, like Monty, also has a special thing about her in terms of her DoWing: If she's Friendly she will accept a war bribe against anyone, even other people she's Friendly with. She's the only AI in the game who will be bribed against someone at Friendly in any situation, but AIs shouldn't be bribed to war against you unless you're already in a war, so that shouldn't be the issue in this case.
AIs don't normally accept bribes when mobilized for war (plotting or already in a war). Even Cathy, the only AI who accepts war bribes against a target she is Friendly with. Not impossible, but unlikely to be an issue. If there is a specific rule to Monty I'm not aware of what it is, but he is the most aggressive AI in terms of base plot chance (2% a turn). And I think he shares that distinction with Ragnar.
The scenarios for an AI to
declare another war when already in one are rare:
-AI plots war over a refused demand, executes DoW later
-AI instantly DoWs over refused demand
-AI dogpiles weak target
-AI is forced into additional war through vassal or alliance mechanics, or the AP holy war
-(obscure) AI was targeting a barb city that you took right before it could. It will attack the city and DoW you in the process. Lain has one of these on video, I think.
By far the most common and likely scenario is a refused demand. AIs handle the retaliatory war actions differently than normal and can either plot them while at war (basically waiting for units to reach your border before DoW), or even hang on to a grudge from a demand in peaceful times and then they get into a war in the interim (likely if some else DoWs them while they plot over demand on you, they don't always drop the grudge-plotting!). You also can't discern whether an AI is plotting over a demand or not if they were already in war mode when it happened, so it can be quite surprising. AIs have two different values - one for the instant DoW for being rebuked, and one for for plotting a sneak attack over a rebuke. Incidentally, Cathy has a value higher than Monty's for plotting a sneak attack if you refuse her.
AIs can also instantly DoW over refused demands, I don't know the details about it other than they don't plot and assemble stacks for it, they just instantly go for it as if they were bribed in. In all cases where it happened to me I bordered the AI and it was stronger than me at the time. Looking in the iVals there are several AIs who apparently can't do this at all (value of 0) which includes Catherine.
AIs will also do the dogpile thing to freely declare wars even when in one already. This requires a certain threshold of power (basically the target must be quite weak in power rating, maybe land controlled as well). This is unlikely to happen unless you are really having a bad game already and generally requires at least one opponent at war with you already.
AIs will get into wars when accepting suzerainty over a vassal already in a war. It behaves much like a bribe (in that it can be blocked by peace treaties with the potential master before it happens). They can accept these arrangements while already in war in some circumstances (shared war against an opponent, they will team up, or if they are strong enough to dogpile the target already). While this doesn't appear to be the case for your example with Cathy
@earthy, it is a potential way a big AI can get into multiple wars at once and it is sufficiently powerful to not be afraid of other players.
A common thread in all methods besides the refused demand thing is that the AI checks power comparisons before determining whether it will commit to a war or not. If it's not afraid of you (threshold is 2x their power rating or something to be immune to normal plotting?) it considers you a valid target, and if you are weak enough (at the dogpile war threshold) it will always consider war with you if it can (attitude) and the opportunity arises. I would
assume it also looks at power comparisons for wars over being rebuked demands as well (so it won't DoW you if you are sufficiently powerful unless you play with Aggressive AI on), but I'm not really sure.