Feature Request/Idea: Have users confirm draft dataset deletion

Currently, an unpublished draft dataset can be deleted with two button presses (cf. #660) with no option to recover a deleted draft dataset. This can lead to loss of work, e.g. when users have already entered and refined metadata over some time. Additionally, when the DOI registered for the draft has already been communicated or added to a paper it is not trivial to get a dataset with the same DOI again.

This has, unfortunately, happened to us and other Dataverse operators in the past (cf. #troubleshooting > Draft DOI - recover record). In our case, users (Depositors) wanted to delete a file and unintentionally deleted the draft dataset instead.

Having an additional safeguard for deletion of draft datasets could help prevent this. One way to solve this could be to prompt the user to type the DOI or DOI suffix of the dataset, similar to how GitLab (and similar platforms) handle deletion of projects.

No files, tests, or entry points are named. Start by locating the draft dataset deletion flow in the frontend and determine how a confirmation requiring the dataset DOI or DOI suffix should fit the existing deletion behavior; done means accidental draft deletion is prevented while intentional deletion remains possible.