FIX cupy import (#5613)

PR fixes an accidental unsafe cupy import from a recent PR. Tests to detect these will be added in 23.12.

Authors:
   - Dante Gama Dessavre (https://github.com/dantegd)

Approvers:
   - William Hicks (https://github.com/wphicks)
   - Ray Douglass (https://github.com/raydouglass)

docs/source/execution_device_interoperability.ipynb

@@ -60,8 +60,8 @@
     "For GPU systems, cuML still follows the [RAPIDS requirements] and nothing has changed for installing it. The cuML package and wheels are universal and can run in both GPU and CPU modes. For installing in CPU systems, similar to other packages it can be installed from conda/mamba with:\n",
     "\n",
     "```bash\n",
-    "mamba install -c rapidsai cuml-cpu=23.10 \n",
-    "# mamba install -c rapidsai-nightly cuml-cpu=23.12 # for nightly builds\n",
+    "mamba install -c rapidsai -c nvidia -c conda-forge cuml-cpu=23.10 \n",
+    "# mamba install -c rapidsai-nightly -c nvidia -c conda-forge cuml-cpu=23.12 # for nightly builds\n",
     "```\n",
     "\n",
     "- cuML 23.10 supports Linux and WSL2 on GPU and CPU systems using conda. \n",

python/cuml/internals/array.py

@@ -14,8 +14,6 @@
 # limitations under the License.
 #
 
-import cupy as cp
-
 import copy
 import operator
 import pickle