PopMan

That's great, thank you for confirming your findings. I wish you luck with your investigation and look forward to an update.

I found and fixed the problem. PopMan 1.3.15 should be as fast as 1.3.1 again...

That's great, thank you for fixing this.
Was the problem related to the introduction of double byte unicode as you suspected?

Yes and no. I used a slow function to convert text to lower case (for case insensitive finding). In one byte ANSI mode that function was fast enough but in Unicode mode it was much slower. I replaced it with a very fast hand crafted function.