Somewhat related to https://bugs.launchpad.net/evergreen/+bug/2045436 but distinct enough that I felt I should make it its own bug.
I feel like I've seen discussion or comments about whether to join the 7XX fields as described in authority.control_set_authority_field to authority.heading_field, but I think it is worth highlighting how a 7XX field interacts with authority.simple_heading_set. As the value in authority.control_set_authority_field.heading_field is NULL be default for all 7XX fields, they are not headings that trigger parsing through authority.extract_headings. However, when extract_headings is called, the MARCXML record is transformed into MADS, in which 7XX headings are considered the same type of heading as a 5XX field, but specifically given the "equivalent" type attribute (<mads:related type="equivalent">).
Thus, extract_headings will still parse and return 7XX fields, but given the function does not currently pay any attention to the type attribute aside from retrieving it, authority.simple_heading is then populated with a version of the heading that is still treated as a 7XX field, as well as one that has been incorrectly returned as a 5XX field.
I see two paths forward: either, give 7XX fields a heading field value so that they parse through extract_headings, or perhaps adjust the type_xpath value for related headings in authority.heading_field so that it does not capture elements with the "equivalent" attribute -- I'm not sure if any 5XX field is ever given that attribute, and ultimately I'd be more inclined to pursue the former option, but I can see the latter being the path-of-least-resistance stopgap for the time being.