We now have Float[np.ndarray, ...] <: np.ndarray. Added basic torch tests. (#68)

This required quite a lot of refactoring! JAX supports virtual subclass registration (its metaclass is ABCMeta) but NumPy does not, so we have to actually subclass `np.ndarray`.
Simple stuff like __base__ hacking fails due to deallocator conflicts.

.github/workflows/release.yml

@@ -33,7 +33,8 @@ jobs:
         with:
             python-version: "3.8"
             test-script: |
-                python -m pip install pytest beartype equinox jaxlib
+                python -m pip install pytest beartype equinox jaxlib cloudpickle
+                python -m pip install torch --extra-index-url https://download.pytorch.org/whl/cpu
                 cp -r ${{ github.workspace }}/test ./test
                 pytest
             pypi-token: ${{ secrets.pypi_token }}

.github/workflows/run_tests.yml

@@ -42,7 +42,8 @@ jobs:
       - name: Install dependencies
         run: |
           python -m pip install --upgrade pip
-          python -m pip install pytest wheel beartype equinox jaxlib
+          python -m pip install pytest wheel beartype equinox jaxlib cloudpickle
+          python -m pip install torch --extra-index-url https://download.pytorch.org/whl/cpu
 
       - name: Checks with pre-commit
         uses: pre-commit/action@v2.0.3