Geonomics

Fixed the nlmpy error I had created by moving the 'module not avail' error inside the only function that uses nlmpy but forgetting to move the import statement there too.

Fixes a bug in species._plot_movement_or_dispersal that accidentally set movement_distr_param2 to the value of movement_distr_param1

Added plot_movement() and plot_dispersal() methods to the Model object, to help with visual design and debugging of sims.

Fixes some minor bugs and better handles when `Species._remove_individuals()` does and does not print out information.

Debugged so that gnx can now row models with rectangular landscapes (checked by plotting [genetic PCA for randomly generated landscape](https://github.com/erthward/geonomics/assets/6105510/61cf9941-a7e1-43ad-ae8a-ce01723fc236), plotting [genetic PCA for a raster-loaded landscape with an msprime-initiated population](https://github.com/erthward/geonomics/assets/6105510/f259e942-6515-43b8-a799-8a735f937a6d), and plotting [phenotypes for raster-loaded landscape with selection](https://github.com/erthward/geonomics/assets/6105510/2e76de1f-0bb5-4a52-95ef-472550e206fa)). Also debugged burn-in for msprime-initiated species, which were still needlessly running one burn-in timestep before dumping and replacing individuals. Also made some other minor tweaks along the way (see commits).

Corrected and clarified documentation for 'rand_genarch' parameter, and added assert statement to make sure that the number of rows in a custom genomic architecture file matches the genome length stipulated in the parameters file. Clarified documentation for 'msprime' parameters block in Species 'init' parameters. Fixed implementation of 'msprime' init parameters so that burn-in is now short-circuited (since the burnt-in population would be entirely replaced by the msprime-derived individuals anyhow).