Cool. Though not sure about ICM implications, raising to win a pot uncontested in an even better result in that case. Playing a 3-bet pot is a bad result, so probably need to play a lot tighter against 3-bets.
I simulated two options:
1: limp or open 4,5bb (though not sure if it's allowed or if the max open is 3,5bb)
2: open 2,5bb
results
1: limp 13,9% open 18,9%
2: open 30,4%
In general in solver land, having more options is better. I'm genuinely surprised that option 2 captured more EV, but it must be because both given lines in option 1 are weaker than the one possible VPIP-line in option 2 at this stack depth.
My conclusion is that while solver likes to have limps in option 1 it's not because limps are great, it's mainly because big opening sizes are very uncomfortable at this stack depth. I'm sure you can capture even more EV by having limps and many opening sizes, but again that is not something I'd recommend.