If you block outright an adversary has reason to try another IP. If you allow the attempt then show a standard "login failed" page they have less information to go on.
Not necessarily. One could have a gov site allowing anyone to view it, but have stricter rules on a /login path, HTTP POST, auth header, or it could have been blocked by some compny-wide safety layer that manages this stuff semi-automatically.
But that's just a speculation.
So the default behavior of a Fortigate is to allow you to apply an access policy to the VPN tunnel itself, which can easily be a geoblock, but the local-in policy where the remote is actually authenticating against the firewall is much harder to change.
Not saying this is a Fortigate or that the federal government didn't change the low effort configuration, but it's certainly not unusual, Fortinet is a huge presence.
Auth providers (like Okta for example) often do the geo-blocking at level 7 -- because if you know the login being used, you can then lock the account that is being accessed from a blocked region.
Remember these are elons are script kiddie hackers, it only occurred to disable the outer firewall, azure ad will independently geoip block all by itself