Made Decision Engine extensible via stevedore

The objective for this is to give the ability to extend the default
set of placement algorithms (i.e. strategies) currently available
in Watcher using Stevedore.

Now, you can add your new strategy as an entry point under the
'[watcher_strategies]' section.

Change-Id: I4aecf629015e41b0389d07e47220333e50bbbe1a

watcher/opts.py

@@ -23,7 +23,6 @@ from watcher.applier.framework import manager_applier
 import watcher.common.messaging.messaging_core
 
 from watcher.decision_engine.framework import manager
-from watcher.decision_engine.framework.strategy import strategy_loader
 from watcher.decision_engine.framework.strategy import strategy_selector
 
 
@@ -38,7 +37,6 @@ def list_opts():
         ('api', watcher.api.app.API_SERVICE_OPTS),
         ('watcher_messaging',
          watcher.common.messaging.messaging_core.WATCHER_MESSAGING_OPTS),
-        ('watcher_strategies', strategy_loader.WATCHER_STRATEGY_OPTS),
         ('watcher_goals', strategy_selector.WATCHER_GOALS_OPTS),
         ('watcher_decision_engine',
          manager.WATCHER_DECISION_ENGINE_OPTS),