🔀 Merge branch 'cuda_gpu' (#52)

Closes #52 Bring in GPU accelerated libraries for rapid ICESat-2 data processing.

README.md

@@ -77,6 +77,11 @@ Then install the python libraries listed in the `pyproject.toml`/`poetry.lock` f
 
     poetry install
 
+If you have a [CUDA](https://en.wikipedia.org/wiki/CUDA)-capable GPU,
+you can also install the optional "cuda" packages to accelerate some calculations.
+
+    poetry install --extras cuda
+
 Finally, double-check that the libraries have been installed.
 
     poetry show

environment.yml

@@ -1,11 +1,14 @@
 name: deepicedrain
 channels:
   - conda-forge
+  - rapidsai-nightly
 dependencies:
+  - rapidsai-nightly::cuml=0.15.0a200817[md5=8e6ede59b1bc5742e21e1e5327173a8b]
   - conda-forge::geos=3.8.1[md5=7282fbe0c036bacad959c03c05a7eb9a]
   - conda-forge/label/dev::gmt=6.2.0.dev0+a1a189c[md5=915cb603613887002b58e09859aaac82]
   - conda-forge::gxx_linux-64=7.5.0[md5=c6afca6a2bfe1ea0ff5566d300460501]
   - conda-forge::nodejs=13.13.0[md5=cf25f8cee7aad7f417980872d34d8fd6]
+  - numba::numba=0.51.0[md5=1151f4a9312696ac4694d82089c92a11]
   - conda-forge::numcodecs=0.6.4[md5=4873a5169bb684b4e120e7bcb287488e]
   - conda-forge::parallel=20200522[md5=8fa06db0afc1e7b1f136277557a4b65e]
   - conda-forge::pip=20.1.1[md5=5e42afe5672d5baf3d9d2260239aa923]