Are you absolutely sure, that no one could have flipped a switch in any way, that has caused this problem somehow on different API´s? The timing this happened (including some taken actions) is more than giving this a fishy smell?

Yes, I'm sure no such switch has been flipped on hive.blog web site or api.hive.blog. BlockTrades operates those two sites, and for security reasons, very few personnel have access to those sites.

For other API nodes, it is up to each operator, of course, but I'm pretty sure none of the API node operators have targeted any users for censorship. And I'm also sure that all the Hive API node operators that I'm aware of use the new code where the centralized blacklists have been removed in favor of decentralized, opt-in lists. So it would require some amount of coding work by the API node operator to block a user in this way.

Condenser itself (the code used for hive.blog) still does have a centralized blacklist, because individual operators may need to hide specific articles to comply with legal regulations (GDPR, DCMA notices, etc), although instances of such blocking are incredibly rare.

Several API nodes are in the process of updating to the new version of hivemind that we just released, which could explain down time/errors on some nodes. And as mentioned already, we do know of one reputation error in the new code that is being worked on now. So we'll probably need to do a new round of upgrades later this week.