Add argparse, deterministic precipitation, image input support for simulation.py

[MightyBOBcnc]

• Results from all scripts other than the machine learning scripts will go into an output subfolder to keep things tidy.
• simulation.py's precipitation randomness has been replaced with seeded randomness in the two parts of the code that were using random precipitation. This makes the output of the simulation deterministic for the same given input seed or image. For now we're just using the current iteration # as the base for the seed and changing it on every iteration to keep things close to the way the previous randomness used to be.
• fbm function also accepts a specific seed for user control of results.
• Introduced argparse to all scripts other than the machine larning scripts to add some features and make things a bit more user-friendly. Run <script_name.py> -h for help on a particular .py file.
• Not all scripts have all arguments but a short rundown is as follows (all arguments are optional except in make_grayscale_image.py and make_hillshaded_image.py):
  -h: Help
  -f: simulation.py can now take an input image as terrain source instead of generating a new fBm every time it runs. If a file is given as input then simulation.py will get its dimensions from the file. Requires a square image.
  -s: A positive integer. This is a seed for the randomness used in plain_old_fbm.py and simulation.py so we can have predictable, repeatable results.
  -o: User-defined name for the output file (without file extension, except in make_grayscale_image.py and make_hillshaded_image.py where user must type .png extension)
  --snapshot: Controls whether simulation.py will use snapshotting.
  --png: Automatically save out a grayscale and hillshaded png of the result to save some work.

[dandrino]

This is a lot of good work, and I like the effort you put into it.

My personal preference for many of these scripts was that they be demos rather than utilities, since I wanted people to go in and change things to see what happened.

However, since I am no longer actively contributing to this project, and you have put in a lot of good effort, I would be willing to add a comment to the top of README.md pointing readers to your fork. Does that sound fine by you?

[MightyBOBcnc]

Hmm, an interesting idea although I'm not sure how to "brand" a link to my fork.

I.E. I'm not really capable of doing major revisions or features related to the simulations as a responsible maintainer or anything like that. I'm still new to programming in general and these demos are using math that's above my head. (I don't remember a thing about imaginary numbers and I never made it as far as Fourier transforms in my schooling. [present-me wishes that past-me had taken calc and trig])

For an example of how I'm already kind of at my knowledge limits, I tried to get simulation.py to operate on a rectangular shape but couldn't figure out how to fix some roadblocks because numpy is like a magic black box to me and I'm not sure how to wrangle the arrays into the proper shape.
ValueError: operands could not be broadcast together with shapes (2,1024,512) (2,512,1024)

I also don't grok, for example, how you plotted the river network images when there doesn't appear to be code for that explicit purpose in the repo.

/images/poisson_disc_sampling.png
/images/delaunay_triangulation.png
/images/river_network_low_inertia.png
/images/river_network_high_inertia.png
/images/river_network_with_volume.png

I know the data is there in their own arrays of course (points, coords, upstream, downstream, volume) but mapping that onto an image eludes me.

So yeah in summation I'm not sure what the best way to frame a link to this fork would be.