Clean up read_deleted support in host aggregates code

tl;dr - this is a cleaner fix for bug #949038

It seems clear to me that all of the DB APIs should not explicitly pass
read_deleted='no' or read_deleted='yes' to model_query without good
reason.

We want to allow callers to specify read_deleted via the context and
that only works if we don't explicitly pass it to model_query().

If we don't explicitly specify it to model_query(), we use the value
from the context which defaults to 'no'.

Given all that, there is no need to support read_deleted to any of the
DB API calls because they should support specifying the flag via the
context. There should also be no need to pass read_deleted='no' because
that is the default.

Really, the only place there should be any mention of read_deleted is
where we want read_deleted='yes' behaviour e.g.

  - In tests, where we want to check the operational_state of an
    aggregate after it has been deleted

  - Where we want to support undeleting an aggregate or aggreate host

Change-Id: I916a8d189a33d7f30838cccb26531a024066ef96

0 files changed, 0 insertions, 0 deletions