Don't test flaky test_complex_series_frame_alignment (#117)

~~I think this test is flaky because this test uses `locals().get` to randomly select a test DataFrame which may not be safe if this test is being run in parallel~~

After iterating on this test in pandas, this test generates DataFrame testing scenarios randomly, but a particular random combination always fails.

xref rapidsai/cudf-private#110

Authors:
  - Matthew Roeschke (https://github.com/mroeschke)

Approvers:
  - Bradley Dice (https://github.com/bdice)

URL: rapidsai/cudf-private#117

python/cudf/cudf/pandas/scripts/run-pandas-tests.sh

@@ -100,7 +100,14 @@ cd pandas-tests/
 
 # TODO: Get a postgres & mysql container set up on the CI
 # test_overwrite_warns unsafely patchs over Series.mean affecting other tests when run in parallel
-PANDAS_CI="1" python -m pytest -p cudf.pandas -m "not single_cpu and not db" -k "not test_overwrite_warns" --durations=50 --import-mode=importlib -o xfail_strict=True ${PYTEST_IGNORES} $@
+# test_complex_series_frame_alignment randomly selects a DataFrames and axis to test but particular random selection(s) always fails
+PANDAS_CI="1" python -m pytest -p cudf.pandas \
+    -m "not single_cpu and not db" \
+    -k "not test_overwrite_warns and not test_complex_series_frame_alignment" \
+    --durations=50 \
+    --import-mode=importlib \
+    -o xfail_strict=True \
+    ${PYTEST_IGNORES} $@
 
 mv *.json ..
 cd ..