The processing function actually have two subroutines, Resample Up and Resample Down, which works by absolutely different methods.
These subroutines are not intended to be complex, so they are not producing best sound quality.
They are necessary for producing non power of two resamplings.
Two more subroutines Sample rate half and Sample rate double have more abilities.
They are only for multiplying or dividing samples amount by 2, and that is the place, where an advanced mathematics are much more easier to implement.
If you need non power of two resample, and care about result sound quality, then do not resample from a rate, which isn't at least twice bigger.
To avoid meaning quality loss you have to use this way :
Use "Sample rate double" until you will get samplerate at least twice bigger than the necessary resulting; at least "plus one" quality method is necessary, and method "ariphmetical average" will not be enough good.
Then you can use "Resample" function to get necessary (at least twice lower) rate, and with it's usage must not appear a big distortions.
For example if you wish to convert from 44100 to 48000 or backward, then duplicate samplerate once (or more), and then convert with Resample from 88200 to 48000.