We can't probably fix your late boost threshold compared to your previous turbo.

Mapping suggestions to lower boost threshold:
Mitsubishi Evos have a lean/retarded map for boost spool assistance only. This is not simple to trigger effectively and safely on standalone ecus. Not impossible but has risk.
You can also map in a spool assist of 100% boost solenoid duty while spooling up. In my experience you only want this to low map figures - say 3psi. The system then goes into a semi run away and you reach your target boost at a lower boost threshold. I don't like the feeling if this is mapped higher than say 3psi as you get the same issue you have with the turbo coming all in too fast for good throttle modulation.

Is the exhaust housing twin scroll?: Expect twin scroll to buy you c.250-500rpm lower boost threshold if it is current single scroll.

I think the main post contributors above all understand that you are hitting your target boost. That does not mean you have good mechanical boost control. At 6psi spring hitting 16-18psi it is poor and won't lend itself to effective boost by TPS mapping. We suggest achieving effective boost by TPS mapping as it will help give you back all the throttle response you want for traction limited autocross.
A good system for a target boost of say 20psi flat would be a 10psi WG spring that gives you 10psi flat out in a high gear, 100%TPS 0% boost DC. With this system control you will much easier achieve a good TPS based boost map and a much more controllable boost/throttle response if desired.

Both the twin scroll and better wastegate priority may be a huge PITA to physically rework - but twin scroll should help boost threshold and separated twin wastegates off the twin manifolds entering the exhaust housing scrolls should provide better wastegate priority - welcome to modified cars.

My 335whp 1.8 is on boost fully by 3000rpm. Twin scroll exhaust housing, early 100% boost DC during spool.

 

Honestly just a simple gtx2860 Gen 2 would suffice. I was able to make just over 400whp with it and the boost threshold was 3500rpm targeting 23psi or so. .64 turbine housing. I think the g25-550 is laggy under threshold the older 2860 really comes on early almost like a GT2560r. No amount of boost controller fiddling will fix the inherent issue with the turbos actual boost threshold

 

I use TPS based open loop boost control. It's great, I get wastegate around 50% throttle, and the second I go WOT it just spools up and makes more boost. It's great for controllability in the corners.

 

So I actually went into the my tune on the G4X and setup a TPS based trim. I haven't been able to test it out aside from on the street, but so far I'm liking it. I'm still running open loop so I basically have my duty cycle set to 100% up until ~40 KPA from target boost where I taper the duty cycle by RPM to hit targets.

I applied a General Purpose trim based on TPS, trimming duty cycle from from 80% down to 0%. So, when I'm at 20% throttle it'll trim roughly 80% duty cycle, and at 80% it'll be trimming closer to 0 (IE, just running the duty cycle from the original open loop map). I don't remember the exact TPS to DC, but basically if I'm not around my target boost the wastegate duty cycle is controlled by the throttle pedal. I thought that might give some weird oscillations, but so far on some spirited drives I've found it to be work pretty well. Previously the car would spool really aggressively without a ton of throttle because the duty cycle was always pegged at 100%, but now I actually have to floor it to get it to go.

At least on the Link, you can also change the trim map by RPM. I did think about changing the TPS vs DC trim before ~4K and after, as the car is usually hitting max boost around 4K. I think you could use that to smooth out the massive torque spike that was the original issue that we were trying to solve.

Again, I haven't been able to test this on track or at autocross and my turbo isn't quite as aggressive, but I'm very excited to try it out when the season starts. Regardless, the car's power actually responds to the throttle pedal now, so that's a win for me even on the street. If anyone's interested I can post up some of the relevant maps from the tune. It might be easier to understand with some pictures/tables.