But I'm not sure if the fee of the replaced transactions that a node relayed wont be removed from its "bucket" and if those will still be counted to its fee estimation after the last replacement is confirmed though.
Doesn't Bitcoin Core look at the amount of time transactions take to confirm with the historical statistics? Ie. Sorting transactions into buckets, analyze fees from a new block by looking at which bucket it is in. If it only considers the fees of transactions which are eventually included into a block, then it would mean that any transactions which are replaced would not matter since it wouldn't be confirmed.
Yes, that is why I need clarification in that matter.
To see if that weird transaction is actually targeting a bug or just a user playing with rbf.
Looks like this is a job of a mining cartel as it tricks Bitcoin Core's fee estimation algorithm.
Mine is just a guess, we can't jump to conclusions because of one wild guess.
You mentioned that it's a "
clear pattern", are there a lot of transactions that has that same behavior?