Issue in recovering perturbation introduced using FFTs

[shyams2]

Commit: 2fac2e0
In the linear theory code when evolving a single mode, we intend to recover the perturbations introduced in the real and imaginary parts by taking FTs of the input distribution function. However, the accuracy of the value recovered is dependant on the resolution considered. This seems to hamper the order of convergence.

For instance, if the input perturbations for the real and imaginary parts are (0.01, 0.02), we find that for N=32 we recover (0.0079, 0.0208) and N = 128, we recover (0.0095, 0.0202).

Current Workaround : The perturbations are setup using the user defined parameters params.pert_real and params.pert_imag. Doing so leads to the expected convergence behaviour

[pavanky]

@ShyamSS-95 @mchandra you can specify the scaling factor as a parameter. you don't need to divide the output separately.