Hard IP bans are the exception and used when a lot of sock puppets are involved operating from the same IP range. Soft IP blocks don't prevent logged in users from editing. On enwiki you can try to get a 12-month IP block exemption: https://en.m.wikipedia.org/wiki/Wikipedia:IP_block_exemption
My strong suspicion is that the blocking is done at a different layer.
Much easier to simply block posts/puts to certain paths from ip ranges in the ingress layer, before they ever reach the application and are authenticated.
That doesn't work because Wikipedia doesn't store your IP forever if you have an account. If they allowed what you described, then you could make a bunch of accounts with your real IP, wait 90 days for the IP you made it from to be forgotten, then vandalize with them all over VPNs, and they couldn't do anything about it other than reactively blocking each account individually.
Then grade accounts by more than one metric. I've my account for years and made plenty of edits but it's treated like some newbie account that may be a bot? Because I value my privacy?
I use Britannica for anything that isn't celebrity/pop culture based now, I thought Wikipedia had killed off "proper" dictionaries but they're going in reverse. I enjoy the irony of that.
I assume the website owner intends to want to show ads or track me for marketing or other data harvesting purchases if they try to do further verification than password + 2FA code. They are forcing CAPTCHAs because they want to make it inconvenient enough such that I disable content blockers on their website.
My decade-old account with thousands of edits and zero vandalization still get hard-blocked behind VPN. I once checked out the exemption process but felt discouraged to make a request (can’t remember why now). It’s pretty stupid.