1. Both services were looking for a direct Harrier/Sea Harrier replacement only.
2. The carriers were originally planned around STOVL operations, with a slow shift to allowing the possibility of future catapult operations only appearing late in the day - the RN had NOT been looking for a US-style attack carrier at all!
This rather confuses the issue of ship size with operating mode and role which is not entirely valid.
It actually went something like this.
Back in the mid-90s the RN was having to think seriously about what capability to replace the CVS with - primarily because the ships would be 30 years old by the end of the noughties. That led to a wide variety of operational analysis studies that ranged from "Do nothing" (ie let the ships and SHAR retire without replacement), through an analogous "replacement in terms of capability" (a CVS-ish CAG and ship), all the way up to a 40 aircraft ship (heavy fixed-wing, light on rotary).
In parallel, the old Director Naval Architecture Future Projects team ( a couple of naval constructors) developed some concept designs based on these OA options, from a 15a/c pure STOVL ship up to a 40a/c CTOL carrier. There were variations on the theme - commercial standard build, two vs three, STOVL vs CTOL, but these were primarily about cost and availability rather than output capability. These were presented at a RINA conference in 1997 and formed the basis of the cost submission for ST(S) 7069.
What the OA demonstrated was that CVS-sized ships did not bring much to the party in the scenarios looked at - most of which involved medium regional conflicts, as opposed to NATO vs Warpac (which was by then defunct). ie Their cost-benefit was marginal. However, if you had 40 cabs aboard you gained a step change in capability and the cost benefit equation was much more valuable. The ship studies showed that if you were sensible about your maintenance requirements and your shipboard systems, you could get the same availability from two big ships as you would for three small ships for broadly the same overall price. The bigger ship also carried less risk in terms of growth in the (as yet undefined) STOVL cab.
That was the basis of the ST(S) submission - two big capable ships of 40000 te that could do Fleet Air Defence, Maritime Strike, Local Air Superiority over a landing area, Land strike etc, plus host a dipper squadron and what was FOAEW/MASC, now Crowsnest. Definitely more than a CVS replacement.
The aircraft that was to fulfill this capability was the Future Carrier-Borne Aircraft (FCBA), for which the various studies assumed a number of options, from the Stovl StrikeFighter (SSF) - which eventually became F35, through a navalised EF2000 operating in STOBAR mode, to an F/A18E/F. ISTR there was even a Harrier SuperVariant as well. THere was a parallel RAF requirement for Future Offensive Air System (FOAS), which was the Tornado replacement.
The operating assumption for the ship was STOVL - the basis for which was essentially a mixture of familiarity, perceived risk in catapult options (we'd got rid of steam, we weren't having a nuclear ship and EMALS back then was veiwed as very high risk) and also a nod to the RAF need to replace its three squadrons of Harriers. I can't actually remember that being an explicit requirement for FCBA, although that did develop in later years (primarily post formation of Joint Force Harrier), when FBCA became FJCA (J being "joint").
Right at the back end of the 90s, after submission of ST(S) 7069 (for the ship), MoD began to look at the options for the FCBA aircraft in more detail, with development of more detailed scenarios, with associated sortie generation requirements and flying programmes. They also wanted to know whether a CTOL ship was invariably going to be bigger (and therefore assumed more expensive) than a STOVL (or STOBAR) ship so they could better cost the overall programme.
At this point, people started to look at deck operation seriously, including getting NAWC in America to do some flightdeck designs, where it became clear that if we wanted to generate lots of sorties, but not have the deck swarming with badgers, chockheads, bombheads and grapes that we couldn't afford, the deck was going to have to get bigger. The 1997 concept designs had been just that - concepts for ROM costing purposes, nothing more, with limited consideration of how the ships would actually work in practice. The next cycle of designs by MoD, BAES and Thales identified that for the bigger ships, 40000 tonnes wasn't going to cut it and you were going to end up significantly north of 50000te even for a STOVL ship. Once you got there, the cost difference between a CTOL and STOVL ship starts getting marginal, although you obviously need to buy the cats and arrester systems. STOBAR was in the same ballpark as STOVL, but generated much less sorties because that mode demands the worst of both worlds in terms of launch and recovery areas, which knocks your safe parking area (crucial for sortie gen) right down. Never mind the comedy attempts to make EF2000 able to see the meatball on a sensible glideslope......
Not long after that, people started to realise that the STOVL aircraft was technically pretty risky and that a hedge against failure was required. This led directly to what was known as the Hybrid design - essentially a large ship, big enough to host two cats and an arrested recovery area, or a STOVL runway,
without drastic modification to the overall design. What that last sentence means is that there would obviously need to be internal arrangement changes for the cat troughs and arrester gear engine room primarily on 2 deck and that the flightdeck strength needed to be designed against the CTOL recovery requirement (irrespective of how she was completed), but you would not have to change the overall dimensions or configuration of the ship. This basic philosophy is what both BAES and Thales submitted their final designs against in 2002 or so prior to the downselect of the Thales design, but appointment of BAES as the prime.
Unfortunately, no-one had updated the Long Term Costings against these larger ships - they were still assumed (by MoD Centre) to be the original ST(S) 7069 cost - which was an unpleasant (but entirely predictable) surprise when the ACA submitted their first project price for the ships and it was £600M over the assumed budget. This led directly to the whole two-year Design Alpha, through Delta exercise where MoD tried to get the cost to match the budget and eventually settled on the current Delta design, which was capable of being completed as either STOVL or CTOL, but - and this is crucial to understand - only if the decision was made at a relatively early stage in build and only if the necessary detailed design work to take the catapults and arrester gear had been completed in time to inform the build. This latter activity was never contracted until the SDSR2010 decision AIUI, by which point, the first ship steelwork was halfway done and the second ship would have had to have been delayed (incurring huge TOBA costs) before design work completed.
Both ships remain convertible if required in future, but it won't be cheap - although it might be cheaper than previously postulated, once the TOBA implications are removed. For example, the electrical generation and distribution software apparently inlcudes modes to account for EMALS loads and EARS inputs.
Tuc's ex IPT bloke was right back in 2003 in that (cost of the cats and arrester themselves excluded) the difference between the operating modes would not have resulted in a new design or a larger ship - ie cost-neutral. However - as ever - that statement needed to be put in context.
Apologies for the essay, but a long way of saying that STOVL does not necessarily preclude the capability of an "attack carrier" in terms of Tacair, although obviously the ASAC/AEW element is a little trickier.