Commit (f2f4f22) was to revert a change I was introducing in this PR. It was a small change, but it shouldn't be included here.

The problem I was attempting to solve: In some cases, when adding an object where subnavigation is exposed, clicking a subnav tab can throw an error if the tab leads to a child of the object currently being added -- because it doesn't exist yet. I was considering turning off subnavigation on the "add" action - but that was shortsighted. There are "add" actions that take place on the subnav tabs themselves, and it's better that these add forms exist in the subnavigation context for continuity and context clues.

Instead, we would do better to keep the subnav but disable a tab's link when this problem occurs.

EDIT: I'm working on a fix for this that aligns with how the fields-nav.inc file was handled (which I missed in this refactoring).

Commit 2eda4f0 fixes subnavigation to align with how it was handled prior to refactoring. This also allows us to use $banners in fields.inc (as we can in columns.inc).