fix wind scaling problem in prsflg==2

[myying]

In mod_random_forcing, the prsflg==2 option scales the wind perturbation magnitude to that specified in the namelist. The scaling analysis use the pressure perturbation magnitude and the pressure wind relationship to compute the expected wind magnitude. But the grid spacing used in the code is minscpx which is not spatially varying. Grid spacing is varying (scpx and scpy) in the computation of pressure gradients dpresx and dpresy, therefore changed to using varying grid spacing in wprsfac computation, so that resulting wind perturbation magnitude won't be affected by varying dx.

[myying]

the specified wind speed amplitude still doesn't match the output exactly. The wprsfac is computed by "pressure_amp / hcorr / fcor /rhoa" the pressure amplitude doesn't exactly drop to zero over the distance "hcorr", roughly half that distance instead. So adding a compensating factor should improve the result, instead of "rhdx" use "0.54rh*dx" as the scale distance.