Missing PAQs are currently filled with 3, this is wrong for different val_ranges

[MitchellAcoustics]

The standard approach to dealing with individual missing PAQs (i.e. someone just forgot to answer e.g. eventful, but answered everything else) has been to fill this with a neutral response. In the standard 5-point Likert scale this is coded as 3, however this is wrong if we have a new range of values (e.g. 0-100 like in SATP). This should either be corrected to just use the mid-point value or choose a different method of dealing with few missing PAQ responses.

[MitchellAcoustics]

Filling with the neutral value is probably wrong to do anyway. Is it possible to just do the ISO calculation while dropping a PAQ? Might need to adjust how the scaling value is calculated.