Thank you for the question @FireForce!
The application currently has no separate app for merging duplicates that already has arisen. The only tools at the users disposal is manual cleanup, moving the data from one record to the other through the tracker capture user interface, and then deleting one of the duplicate records.
We are however working on an app and services for making this happen, and you can follow and interact with this work in this Jira issue: [DHIS2-2107] - Jira
The mecanisms that is already in place in tracker is mainly to avoid duplicates in the first place. While registering a new tracked entity instance, the tracker does its best to search for and alert the user of any possible duplicates. I would take a look and make sure that both the program, and especially the tracked entity type is set up with a sensible set of tracked entity attributes that is searchable. This is extra important when no unique ids exists for the use case.
Best regards,
Markus