Freeze top-level packages at arbitrary existing versions

We have not updated this image in ages, and when we just
bumped up a couple of
packages (#3),
it actually bumped up a *lot* of other packages to newer versions.
This will break a lot of users in the middle of a semester, which
is problematic.

Ideally, we should have 'proper' pinning (like in
https://github.com/CryoInTheCloud/hub-image), but given the lack
of that, I have just specifically pinned the versions to what we
currently have while we work on a longer term solution.

requirements.txt

@@ -1,22 +1,22 @@
-numpy
-scipy
-matplotlib
-pandas
-seaborn
-sympy
-networkx
-scikit-learn
-beautifulsoup4
-statsmodels
-nltk
-spacy
-geopandas
-altair
+numpy==1.21.6
+scipy==1.7.3
+matplotlib==3.5.2
+pandas==1.3.5
+seaborn==0.11.2
+sympy==1.10.1
+networkx==2.6.3
+scikit-learn==1.0.2
+beautifulsoup4==4.11.1
+statsmodels==0.13.2
+nltk==3.7
+spacy==3.3.1
+geopandas==0.10.2
+altair==4.2.0
 # From https://github.com/utoronto-2i2c/jupyterhub-deploy/issues/29
-astroquery
-astropy
-galpy
-catalystcoop.pudl
+astroquery==0.4.6
+astropy==4.3.1
+galpy==1.7.2
+catalystcoop.pudl==0.3.2
 dask==2020.12.0
 distributed==2020.12.0
 dask-gateway==0.9.0
@@ -33,10 +33,10 @@ jupyter-rsession-proxy==1.2
 jupyter-tree-download==1.0.1
 ipywidgets==7.5.1
 otter-grader==2.2.6
-datascience
-geojson
-folium
-geopy
+datascience==0.17.5
+geojson==2.5.0
+folium==0.12.1.post1
+geopy==2.2.0
 # JupyterLab extensions
 jupyterlab_widgets==1.0.*
 jupytext==1.10.*