use `unique()` on task_names input in create_task_plan

I ran into this and thought something was funky w/ my build, since my task makefiles were taking so long to build. I was using a huge task_names vector that contained many duplicates (I had simply no realized I didn't run `unique()` on it. Somehow, `create_task_plan` is smart enough to only end up with the unique targets, but it churns for awhile with all of the extra inputs.

This seems like a small issue to deal with - potentially warn when there are duplicate task names and use `unique()` on them?

Start at the create_task_plan entry point and inspect its existing tests or callers to understand how task_names is processed. Confirm the behavior with a duplicated task_names vector, then define done as avoiding unnecessary work for duplicate names and, if retained, covering the duplicate-input behavior with a test.

Written by the indexing model from the issue text.