use mgcv to allow for exchangeable spline bases

[julia-wrobel]

Using the mgcv package to create spline bases for the mean and eigenfunctions. This will allow the user to easily swap out different types of basis functions. Making this change will require a few considerations:

• Basis can be generated using Theta_phi = smoothCon(s(hinv_tstar, k=Kt, bs="bs"), data=data.frame(hinv_tstar=hinv_tstar))[[1]]$X
• More knots and basis functions are generated, the total number of knots will be Kt + 4
• Several boundary knots are created in order to have better behavior of the tails of the function.
• Basis for warping functions, Theta_h, will still be generated bs()

[jeff-goldsmith]

Just so I'm clear -- the goal is to use mgcv to generate the basis functions but not to implement the smoothing for the mean and eigenfunctions, correct?