implement a sql catalog backend

[Joseph Heck]

implement a sql based (sqlalchemy) backend for the Catalog service

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/4464
Committed: http://github.com/openstack/keystone/commit/37d223ecdb392f3b46079418a7b82398afca2128
Submitter: Jenkins
Branch: master

commit 37d223ecdb392f3b46079418a7b82398afca2128
Author: Adam Gandelman <email address hidden>
Date: Thu Feb 23 13:16:00 2012 -0800

    Implement a Catalog SQL backend

    This adds a catalog SQL backend. Makes use of 3 tables: endpoint,
    service and service_endpoint_association. Services and endpoints are
    joined via the association table. New sqlalchemy models have been defined
    in keystone/catalog/backends/sql.py and are imported during the initial
    migration (v001).

    Configuring the service catalog is possible with changes to
    python-keystoneclient. I will be proposing a merge for that and I'll
    update this commit msg with a link to its review. With those client
    changes, admins can now create and delete endpoints that are associated
    with existing services. Existing service commands on the client-side work
    as expected against this new backend. This driver's get_catalog method
    properly translates existing services, endpoints and relatoins into something
    consumable by keystone non-admin users / clients.

    Update: Some cleanup as per bcwaldon's suggestions

    Update: Match functionality of existing catalog backend by returning
            IDs instead of sql objects for list_services() and list_endpoints()

    Update: pep8 fixes

    Update (1/2): Remove legacy OS-KSADM stuff

    Update (2/2): Remove ServiceEndpointAssociation table/model
                  in favor of a FK, endpoint.service_id -> service.id

    Resolves bug 928053

    Change-Id: Icc11889920744c36255f06356744cb247d79f4aa