Try "processes" scheduler in gallery example

When running the plot_haar_extraction_selection_classification.py
directly it fails with the error attached at the end. However, executing
the file via sphinx gallery or its content in an IPython console seems
to work. So test it in the CI.

RuntimeError:
        An attempt has been made to start a new process before the
        current process has finished its bootstrapping phase.

        This probably means that you are not using fork to start your
        child processes and you have forgotten to use the proper idiom
        in the main module:

            if __name__ == '__main__':
                freeze_support()
                ...

        The "freeze_support()" line can be omitted if the program
        is not going to be frozen to produce an executable.

        To fix this issue, refer to the "Safe importing of main module"
        section in https://docs.python.org/3/library/multiprocessing.html

doc/examples/applications/plot_haar_extraction_selection_classification.py

@@ -25,9 +25,8 @@
 
 """
 from time import time
-from functools import partial
-from multiprocessing.pool import Pool
 
+import dask
 import numpy as np
 import matplotlib.pyplot as plt
 
@@ -66,11 +65,13 @@ def extract_feature_image(img, feature_type, feature_coord=None):
 # To speed up the example, extract the two types of features only
 feature_types = ['type-2-x', 'type-2-y']
 
-# Use multiprocessing pool to compute tasks in parallel
-worker = partial(extract_feature_image, feature_type=feature_types)
+# Build a computation graph using Dask. This allows the use of multiple
+# CPU cores later during the actual computation
+X = [dask.delayed(extract_feature_image)(img, feature_types) for img in images]
+X = dask.delayed(np.stack)(X)
+# Compute the result
 t_start = time()
-X = [x for x in Pool().map(worker, images)]
-X = np.stack(X)
+X = X.compute(scheduler="processes")
 time_full_feature_comp = time() - t_start
 
 # Label images (100 faces and 100 non-faces)
@@ -137,15 +138,16 @@ def extract_feature_image(img, feature_type, feature_coord=None):
 # but we would like to emphasize the usage of `feature_coord` and `feature_type`
 # to recompute a subset of desired features.
 
-# Use multiprocessing pool to compute tasks in parallel
-worker = partial(
-    extract_feature_image,
-    feature_type=feature_type_sel,
-    feature_coord=feature_coord_sel
-)
+# Build a computation graph using Dask. This allows the use of multiple
+# CPU cores later during the actual computation
+X = [
+    dask.delayed(extract_feature_image)(img, feature_type_sel, feature_coord_sel)
+    for img in images
+]
+X = dask.delayed(np.stack)(X)
+# Compute the result
 t_start = time()
-X = [x for x in Pool().map(worker, images)]
-X = np.stack(X)
+X = X.compute(scheduler="processes")
 time_subs_feature_comp = time() - t_start
 
 y = np.array([1] * 100 + [0] * 100)