Host Manager should use monitor metric objects

[Sudipta Biswas]

Per bug 1468012 we changed the monitor metric reporting to use versioned Monitor Metric objects instead of plain old dictionaries.

This bug is being filed to address a code refactoring needed inside nova/scheduler/host_manager.py - in the _update_metrics_from_compute_node to use the monitor metric object instead of the current non object implementation.

[Matt Riedemann (mriedem)]

Simple code cleanup doesn't warrant a bug IMO.

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/213317
Committed: https://git.openstack.org/cgit/openstack/nova/commit/?id=ae7dab9975bcbe3bb40cb9723b0deaed985b904c
Submitter: Jenkins
Branch: master

commit ae7dab9975bcbe3bb40cb9723b0deaed985b904c
Author: Sudipta Biswas <email address hidden>
Date: Sat Aug 15 01:15:08 2015 +0530

    Fix scheduler code to use monitor metric objects

    Per bug 1468012 we changed the monitor metric reporting
    to use versioned Monitor Metric objects instead of plain
    old dictionaries.

    This fix addresses a code refactoring needed inside
    nova/scheduler/host_manager.py - in the
    _update_metrics_from_compute_node to use the monitor metric object
    instead of the current non object implementation.

    Further, it fixes the metrics_filter.
    The metrics_filter was expecting a dict of metrics but now
    it has to adapt to the MonitorMetricList object instead.

    Closes-bug: 1485082
    Change-Id: I4dfbea27ce6c3eecc1a8658b1f9dc0feb2298705