Bulk import does not delete nested metadata properly

[olli-gold]

Describe the bug
When items are imported with the bulk-import script nested metadata groups can be set (for instance a link consisting of an URL and a title, cf. https://github.com/4Science/DSpace/blob/dspace-cris-7/dspace/config/submission-forms.xml#L365). Such metadata needs to be put into an extra tab in the excel import file (the website for a person for instance should be put into a tab with the name oairecerif.identifier.url). That works fine, unless I try to delete such a website via the bulk import without wanting to delete the entire person item. I tried to do that by setting an empty row into the extra tab in the import file (only referencing the parent ID). Doing that will delete the Website, but it will set a Placeholder into the field for the title (i.e. crisrp.site.title).

This could not be a problem at all, but it causes trouble on the item page: although there is no website any more, an empty row is displayed on the item page.

To Reproduce
Prerequisite: you have an item page for persons, which shows the metadata field oairecerif.identifier.url (this is the case for an out-of-the-box DSpace-CRIS).

1. Bulk-import this Excelfile into your system (into the person collection):
   create_websitetestitem.xls
2. Check the newly created person.
3. Bulk-import this Excelfile into your system (be sure to insert the correct ID of the newly created person from step 1 before executing this):
   update_websitetestitem.xls
4. Check the now modified person.

Expected behavior
I would expect, that the website is deleted entirely with the update.