Oh no, you are most definitely talking about anti-personnel cluster (sub)munitions: "but rather to increase the CEP of the terminal warhead and allow it to tear up a carrier's topside flight deck which is a very soft target with the island filled with radars and people, and the flight deck with arrestor gear, catapults". Anti-armor cluster submunitions are much larger and fewer in number and would not appreciably increase the probability of a hit (BTW "increasing CEP" literally means you are worsening the accuracy of the ASBM).
When I say "increasing CEP" what I mean is that a cluster submunition would allow a cluster warhead to have a greater CEP (i.e.: yes, to be less accurate) while also still allowing a portion of the cluster warhead's dispersed submunitions to hit the carrier, compared to a unitary warhead.
We are talking (more or less) single digits for anti-armor vs double or triple digits for anti-personnel. Anti-armor submunitions are also not intended to attack "very soft" targets. If they kill people or radars, it's because they randomly happened to land right on top of them, whereas an anti-personnel submunition could still conceivably kill a person or cripple a radar from a few dozen feet away. So you were definitely talking about anti-personnel cluster bombs, that much is clear.
As for the number of munitions that an anti armour cluster munition can carry, well the CBU-100/mk-20 rockeeye which weighs about 220kg carries 247 unguided Mk-118 anti tank bomblets -- not exactly single digits in terms of number of bomblets if that's what you're referring to by this part: "We are talking (more or less) single digits for anti-armor vs double or triple digits for anti-personnel".
Now, I want to make it clear that I'm
not saying that my hypothetical AShBM cluster munition warhead will carry Mk-118 style bomblets, nor am I saying Mk-118 bomblets will be suitable for striking a carrier's topside deck. The CBU-100 is only brought up as an example for a cluster bomb with anti armour bomblets which has a fairly large number of submunitions within it and its nature as an area-effect weapon, as reply to your "single digits for anti-armour" part.
So in conclusion, no, I definitely was not talking about any specific kind of cluster bomb, and I leave open the field for anti personnel, or anti armour submunitions both as submunition types, or another potential submunition type that might be specialized for the AShBM (who knows).
If you want to do significant damage to the flight deck itself, you have to go with anti-armor submunitions because you are simply not even going to stipple a carrier's thick flight deck armor with grenade-sized fragmentation devices. They would also either have to be modified or be specially-designed "anti-deck" submunitions rather than anti-armor since pretty much all modern anti-armor submunitions are programmed to look for and attack tanks and IFVs with a shaped-charge warhead (incidentally with very little splash damage). Personally I would go with anti-personnel submunitions. These would increase hit probability and create significant and permanent losses in crew, aircraft and other hardware and cause fires that would take hours to days to put out a la Forrestal and Enterprise fires writ large. Meanwhile a hole the size of an anti-armor submunition punched into a flight deck may or may not have any appreciable effect on flight ops and could possibly be patched over in short order.
Like I said, I deliberately did not commit to either an anti personnel or anti armour style submunition because I wasn't sure if either one of them (or neither) would be most optimal for the mission I was envisioning.
I also definitely did not suggest that any cluster munition warhead (whether it be anti armour or otherwise) would directly adapt an anti armour submunition like the sort of targeted guided submunitions to target AFVs with EFPs like the CBU-97's skeet submunitions, and when I talk about the possibility of anti-armour submunitions I'm not thinking about CBU-97 type guided skeet EFPs.
As for what kind of warheads and what kind of targets to optimize cluster munitions for -- IMO the most likely "available" targets on the carrier's topside include the island+associated radars, the flight mission related hardware like catapults and arrestor gear, and finally the flight deck itself (if the ability to afflict fairly widespread damage is able to be committed in the last case).
One reason I do not consider the carrier's flight deck personnel and the aircraft to be an optimal target is because I expect one of the aircraft carrier's primary countermeasures to an AShBM to be to stow as many of their aircraft and personnel below decks in the hangar and within the ship respectively, once they receive early warning of an AShBM launch.
OTOH, things like the island, the catapults, the arrestor gear, and the flight deck itself cannot be hidden and protected under a layer of something else. They are all permanently exposed, and if any of them are significantly damaged, it will mission kill the carrier's ability to effectively carry out its mission of launching and recovering aircraft.
Whether an anti personnel, anti armour, or different type of sub munition would be most optimal to fulfill that role is not an answer I have, but that is the kind of target set that I think a submunition AShBM warhead should be looking to be designed for.