Resolve #494: Reduce duplicate and unnecessary database requests

[seav]

Proposed changes in this pull request

• Resolve Weed out duplicate database requests in templates #494:
  • Use aggregation and annotation (select_related(), prefetch_related(), and annotate()) to combine a list query and the subsequent individual list item queries into just 1 database access whenever possible. (The attachment count of resources is not optimized because it seems aggregation is impossible (?) due to ContentObject’s generic relations.)
  • Use only() or defer() to only retrieve relevant fields or to avoid retrieving the attributes field when it is not needed. Also add get_queryset() managers for the SpatialUnit, Party, and TenureRelationship to defer the models’ attributes field by default. Note: SpatialRelationshipManager was moved from models.py to a new managers.py file.
  • Add simple caching when objects, projects, or organizations are first retrieved. Add the CacheObjectMixin for views that do not have an explicit get_object() method. And switch a single-object view from using get_queryset() to instead declaring the model and leveraging CacheObjectMixin.
  • Improve SpatialQuerySetMixin.get_queryset() to either return all project locations or all excluding the current location for DRY purposes.
  • Update the is_administrator() methods of OrgAdminCheckMixin and ProjectAdminCheckMixin:
    • to inherit from SuperUserCheckMixin instead of having the superuser check code again.
    • to check that the user is an org/project administrator by trying to search for their OA or PM roles in OrganizationRole and ProjectRole instead of constructing a list of org admins or project admins and then checking if the user is in the list.
  • Leverage SuperUserCheckMixin, OrgAdminCheckMixin, and ProjectAdminCheckMixin as needed instead of having repeated code.
  • Remove redundant get_context_data() declarations or code if they can be inherited from a mixin.
  • Simplify reverse() calls in views to reuse already available URL kwargs instead of retrieving organization and project objects.
  • Update templates to use already existing and new context variables.
  • Update unit tests as needed. Special note: Because of the deferred fields, some tests were switched from using refresh_from_db() to retrieving a fresh instance from the database.
• Other cleanup and improvements:
  • Remove a redundant conditional in a template.
  • Switch around the username and full name of the resource contributor in the resource widget.
  • Switched some Python 2-style super() calls to Python 3.

Limitations

• The API was left untouched (unless they used common mixins). This should not be a big issue because the API is generally conservative with database requests by design.
• Duplicate database requests that are traced as being issued inside django-tutelary or django-jsonattrs are left alone.
• See the limitation above regarding the attachment count of resources. I would recommend ditching this count as this will slow down the website when we have tons of resources.

When should this PR be merged

For Sprint 10.

Risks

None foreseen. The updated code should be functionally the same to the current code (except for the username/full name switch-around in the resources widget).

Follow up actions

None. (But, resolving #572 can further cut down on database requests but that issue needs discussion.)