My main gripe with UCAVs is that in non-permissive environments they will be jammed to death.
The loyal wingman concept minimizes that issue because the UCAVs would be operating closer to the command node.
But I don't think you would be able to control the loyal wingman aircraft easily in a single seater. You will need a twin-seat aircraft with a drone operator IMHO. Neither the F-22 nor the F-35 are twin-seat.
I think that as long as China has 4th gens with high power AESA radars, IRST, and HMDs for target acquisition plus the requisite weapons with an adequate communication network and ECM capabilities the stealth aircraft won't be as effective as claimed. At least the F-22 and F-35 in its current configuration won't be. Even less so if the Chinese have a comprehensive radar network with long-wave radar that is stealth resistant.
The Chinese will mainly need their own stealth aircraft to operate outside the envelope of their A2/AD bubble.
That is basically what the Russians are doing and I think is a lot more cost-effective than masses of 5th generation aircraft.
Sure it would be nice to have a twin-seat J-20 attack variant or a single-engine single-seat lower cost stealth fighter but I think those have a lower priority.
Couple of points
Spread spectrum comms is standard, and is very difficult to detect or jam.
But let's say drone datalinks will be under attack.
That means you need autonomous UCAV swarms with the authority to detect, identify and shoot without outside C&C.
In such scenarios, how many manned stealth fighters would you need?
As for the effectiveness of non stealthy 4th Gen fighters, they can be detected by long range UHF or L-Band radar.
Then engaged by X-band air to air missiles.
So I struggle to see how 4th gen fighters can be effective against a stealth fighter which is immune to radar guided missiles.