Hi,
Basically this sounds like it could work using metric_relabel_configs.
However, it also sounds like an anti-pattern, both the "let's federate
(almost) everything" part and the "let's find a workaround for a
specific team".
Maybe you should look into empowering the team to work with "their"
Prometheus instance and try to keep the federated metrics at a minimum
(meta monitoring, job-level stuff)?
Kind regards,
Christian