assigner: major work and new arguments

v.0.2.7 you need to update stackr to v.0.2.7 to appreciate this new version of assigner. updated assignment_ngs with the separate stackr module to simplify the function. new arguments in assignment_ngs. The assignment using dapc can now use the optimized alpha score adegenet.dapc.opt == "optim.a.score" or the cross-validation adegenet.dapc.opt == "xval". This is useful for fine tuning the trade-off between power of discrimination and over-fitting (for stability of group membership probabilities). Cross validation with adegenet.dapc.opt == "xval" doesn't work with missing data, so it's only available with imputed data (i.e. imputation.method == "rf" or "max"). With non imputed data or the default: the optimized alpha-score is used (adegenet.dapc.opt == "optim.a.score"). When using adegenet.dapc.opt == "xval", 2 new arguments are available: (1) adegenet.n.rep and (2) adegenet.training. See documentation for details. removed arguments in assignment_ngs. Removed the pop.id.start and pop.id.end arguments that were confusing people. For those used to these arguments, they are now recycled in the new function individuals2strata in stackr. The strata file created by this function can be used with the strata argument in assignment_ngs. 2 modified arguments in assignment_ngs: (1) gsi_sim.filename is now filename; and (2) if you didn't use the imputation argument, replace imputation.method = FALSE to imputation.method = NULL or leave the argument missing. simplified sections of codes in assignment_ngs that dealt with strata, pop.levels and pop.labels. new function: write_gsi_sim. Write a gsi_sim file from a data frame (wide or long/tidy). Used internally in assigner and might be of interest for users