I have taken a quick look at your file and the WHole2 had a comma at the end so it isn't being treated as a numeric;
I have to get a flight at 6am tomorrow but I think if you update your input file you'll be getting closer to what you want as the deconvolution is now working for Whole2. I would caution you that some of your Diss profiles 'cross' e.g. Jam and 400
Also if 900 is indeed your target then you will be extrapolating a long way from your other test forms...
Lastly when you make your prediction you should be using the same Dissolution model as in your correlation, currently it is set to Hill whereas you original dissolution used Weibull