[BUG] Small fp precision failure in linear regression doctest test

[dantegd]

Describe the bug A couple of nightly runs, like https://gpuci.gpuopenanalytics.com/job/rapidsai/job/gpuci/job/cuml/job/branches/job/cuml-gpu-arch-branch-22.08/CUDA=11.5,GPU_LABEL=gpu-t4/39/testReport/cuml.tests/test_doctest/test_docstring_LinearRegression_/, have shown the following pytest failure:

E         File "/opt/conda/envs/rapids/lib/python3.8/site-packages/cuml/linear_model/linear_regression.cpython-38-x86_64-linux-gnu.so", line ?, in LinearRegression
E         Failed example:
E             print(preds)
E         Expected:
E             0   15.999...
E             1   14.999...
E             dtype: float32
E         Got:
E             0    16.000000
E             1    14.999999
E             dtype: float32