It appears that the limitations imposed on time-series collections are temporary because in the upcoming version MongoDB 5.1, we can perform some delete and update operations. I’m guessing time-series might support upsert and the unique property on secondary indexes in future releases.
I came up with my own method to ensure data is never retrieved from an API and accidentally inserted twice, which was my original problem.
I’m still learning the aggregation pipeline, but @sergeimeza way of creating a query that ignores duplicates is probably the best solution for now.
Thanks for all the help.