Add a baseline set of _MultiCall performance tests

This begins an effort to incorporate run-time speed tests using
`pytest-benchmark`. This initial test set audits the `_MultiCall`
loop with hookimpls, hookwrappers and the combination of both.
The intention is to eventually have a reliable set of tests which
enable making core component modifications without disrupting
performance as per #37.

testing/test_multicall.py

@@ -22,7 +22,7 @@ def MC(methods, kwargs, firstresult=False):
     for method in methods:
         f = HookImpl(None, "<temp>", method, method.example_impl)
         hookfuncs.append(f)
-    return _MultiCall(hookfuncs, kwargs, {"firstresult": firstresult})
+    return _MultiCall(hookfuncs, kwargs, firstresult=firstresult)
 
 
 def test_call_passing():
@@ -197,3 +197,40 @@ def m2():
     with pytest.raises(exc):
         MC([m2, m1], {}).execute()
     assert l == ["m1 init", "m1 finish"]
+
+
+@hookimpl(hookwrapper=True)
+def m1(arg1, arg2, arg3):
+    yield
+
+
+@hookimpl
+def m2(arg1, arg2, arg3):
+    return arg1, arg2, arg3
+
+
+@hookimpl(hookwrapper=True)
+def w1(arg1, arg2, arg3):
+    yield
+
+
+@hookimpl(hookwrapper=True)
+def w2(arg1, arg2, arg3):
+    yield
+
+
+def inner_exec(methods):
+    return MC(methods, {'arg1': 1, 'arg2': 2, 'arg3': 3}).execute()
+
+@pytest.mark.benchmark
+def test_hookimpls_speed(benchmark):
+    benchmark(inner_exec, [m1, m2])
+
+
+@pytest.mark.benchmark
+def test_hookwrapper_speed(benchmark):
+    benchmark(inner_exec, [w1, w2])
+
+@pytest.mark.benchmark
+def test_impl_and_wrapper_speed(benchmark):
+    benchmark(inner_exec, [m1, m2, w1, w2])