I have to agree with @kutzi that having published plugins known to cause abysmal performance problems is not a good idea. Lots of people will just blindly install stuff and not read the warnings. Better to unconditionally delete bad code.
If there is a real use case for mail address resolvers that infer information from build records, then they should be written carefully to have reasonable performance. For example:
- require the user to specify a sample job or jobs, rather than searching every job in the Jenkins installation
- inspect only up to some maximum number of build records per job (with lazy loading this is especially critical)
- cache information obtained from such searches, e.g. using the Database plugin, so you have only a one-time hit