"Use an agent" isn't an appropriate response for jobs that must be run on the master. For example, a backup job that backs up the master. For example, a monitoring job that checks for some problem on the master–configuration change, out of disk space, etc. For example, a build job that uses a plugin with a bug where it won't run on the agent.
Once you allow these types of jobs to run on the master, some user can configure their build to run there for a while before you notice and make them move it. (This, of course, turns into disk space that would never be reused once you get them running on the agent.)
Plus multibranch pipeline jobs use a workspace when there are shared libraries to load from SCM and some other situations.
It works really well to move these jobs to a location on a different volume of the file system (as explained above)
Finally, if there is no reason to move the workspace for multibranch pipeline jobs, there is also no reason to move other job types. All the considerations are the same except the multibranch pipeline jobs create more workspaces because there can be one for every branch. The other types of jobs don't create and discard workspaces as the branches are created and discarded.