I don't really understand the changes here.

You're testing the db status in tests for the chronik handler, the events handler (is that different ?) and the main app (same ?). you're duplicating the test intent over these tests and it's why it's causing all these duplicates.
Keep the things self contained and avoid yourself the exponential maintenance cost.

Ultimately, these functions are supposed to update the database. I think the best way to avoid the expanding maintenance cost is to require the top level functions to "return a callback" ... then the unit test for example of main.js is just looking "did it call the expected function" and not "did the entire chain of updating the database work out as we've expected." This tho also gets complicated, bc then we are loading up all these parent functions with numerous callbacks.

Root cause of the diff is I need to show that handleBlockConnected clears out pendingAliases that have confirmed. In writing this unit test, I noticed that it failed because the unprocessConfirmedTxs were not correct, due to the server state issue. So, rather than just add one test where it's "fixed" to show this pending alias cleared behavior, fixing it everywhere.

I'm open to alt approaches but imo the complication of passing callbacks is a little worse than the current situation, where we are perhaps over-testing.