I put this in mainly for QRDA imports but could be possible with CCDA.
What can happen is documents could be received in a batch import that may use the same encounter with additional content across two different documents.
Very low chance this will occur with CCDA or internal openemr however with QRDA depending on what measures are imported, one could have the same encounter across two documents with different content depending on what content the measure requires to calculate the measure.
I seems sort of silly however ONC requires the capability and at the time I finally reasoned it is a potential gotcha so there it is.
In the end though, I automatically resolve regardless of which button is used. I left it in for testing to help keep the import workflow straight as a procedure for the person running the test.