Thank you all for your contributions and views. To summarise:
It is feasible to train to convert / become competent on a plane entirely/primarily in a simulator.
Pilots need to fly real hours to remain current - but these don't necessarily need to be undertaken on the advanced asset.
Some things cannot be replicated in a simulator (primarily sensations such as 'g' or the 'fear' factor which can only exist when flying a real mission). But flying real missions risks being observed and there are significant issues with replicating credible /realistic threat environments in the air. Regards the latter, if a real mission isn't realistic 'what's the point?' (from an operational / training perspective).
Projecting forwards, there are several reasons not to fly the advanced assets very often. It's expensive, it is risky from a security perspective and it's hard to replicate missions. So these assets could sit around in pristine condition and flown a few hours a month. The pilots need to keep up their hours (this could be on a cheaper asset - Aeralis?), train in simulators and get a few hours/months in the main asset (to compare sim experience with real life).
The implications for industry /MoD which I alluded to in my reply to BV thus seem very real. Does anyone else have more to contribute?
Thank you, Scrimshankers