Use pytest instead of py.test per upstream recommendation, #dropthedot

http://blog.pytest.org/2016/whats-new-in-pytest-30/
https://twitter.com/hashtag/dropthedot

xdist/dsession.py

@@ -18,7 +18,7 @@ class Interrupted(KeyboardInterrupt):
 
 
 class DSession(object):
-    """A py.test plugin which runs a distributed test session
+    """A pytest plugin which runs a distributed test session
 
     At the beginning of the test session this creates a NodeManager
     instance which creates and starts all nodes.  Nodes then emit
@@ -26,7 +26,7 @@ class DSession(object):
     methods.
 
     Once a node is started it will automatically start running the
-    py.test mainloop with some custom hooks.  This means a node
+    pytest mainloop with some custom hooks.  This means a node
     automatically starts collecting tests.  Once tests are collected
     it will wait for instructions.
     """

xdist/looponfail.py

@@ -1,5 +1,5 @@
 """
-    Implement -f aka looponfailing for py.test.
+    Implement -f aka looponfailing for pytest.
 
     NOTE that we try to avoid loading and depending on application modules
     within the controlling process (the one that starts repeatedly test

xdist/workermanage.py

@@ -184,7 +184,7 @@ class HostRSync(execnet.RSync):
 
 
 def make_reltoroot(roots, args):
-    # XXX introduce/use public API for splitting py.test args
+    # XXX introduce/use public API for splitting pytest args
     splitcode = "::"
     result = []
     for arg in args: