On CI, just let pip resolve most dependencies.

Author: anntzer

.appveyor.yml

@@ -86,9 +86,8 @@ install:
   #
   # n.b. pandas is pinned to <0.21 due to issue 9610
   - conda create -q -n test-environment python=%PYTHON_VERSION%
-    pip setuptools numpy python-dateutil freetype=2.6 msinttypes "tk=8.5"
-    pyparsing pytz tornado "libpng>=1.6.21,<1.7" "zlib=1.2" "cycler>=0.10"
-    mock sphinx "pandas<0.21.0"
+    msinttypes freetype=2.6 "libpng>=1.6.21,<1.7" zlib=1.2 tk=8.5
+    pip setuptools numpy mock "pandas<0.21.0" sphinx tornado
   - activate test-environment
   - echo %PYTHON_VERSION% %TARGET_ARCH%
   - if %PYTHON_VERSION% == 2.7 conda install -q backports.functools_lru_cache

.circleci/config.yml

@@ -59,7 +59,7 @@ pip-run:  &pip-install
 deps-run: &deps-install
   name: Install Python dependencies
   command: |
-    pip install --user python-dateutil numpy${NUMPY_VERSION} pyparsing!=2.1.6 cycler codecov coverage sphinx pillow
+    pip install --user numpy${NUMPY_VERSION} codecov coverage
     pip install --user -r doc-requirements.txt
 
 mpl-run: &mpl-install

.travis.yml

@@ -116,10 +116,7 @@ install:
         $PANDAS \
         codecov \
         coverage \
-        cycler \
         pillow \
-        pyparsing!=2.1.6 \
-        python-dateutil \
         sphinx
     # GUI toolkits are pip-installable only for some versions of Python so
     # don't fail if we can't install them.  Make it easier to check whether the