"Refine > Associate Proteins" should not remove peptides from the document.
Unmapped peptides which were originally inside of a Peptide Group should remain in the same Peptide Group that they were in before.
Unmapped peptides which were originally inside of a Protein or Protein group should be moved to a Peptide Group called something like "Unmapped Peptides".

What about decoy peptides?

Decoy peptides should remain in the Peptide Group that they already belong to.
Since "Associate Proteins" does not result in any peptides being removed or added to the document, the decoy peptides that were already in the document are still valid decoys.

I'm adding an option to keep peptides. Some users may want to discard unmapped peptides and it's simple to enable both. Also, it will always be possible for associate proteins to remove peptides by removing proteins due to the min peptide count filter (and that protein has some peptides that only occur in that protein).

What about regenerating decoys if a decoy list exists?