1. Since unicode-data is now a large package we should provide a prominent warning (in README/package description) that this may be a heavier dependency, taking more time to compile and more disk space. Users may want to consider depending on individual packages esp unicode-data-core for lightweight use.
2. We should proactively inform the existing users of unicode-data (using reverse dependencies, maybe raise issues in their repos) that they should depend on unicode-data-core instead, otherwise this may come as a surprise to them.

Note: This is currently based on #92 and will need to be rebased after #92 is merged.

You could choose wip/Unicode_15.0.0 as the base branch for this PR.

I agree with your proposals. I think there is no rush, so we can wait that users first update to 0.4 (#92). Then it would be a good time to remove deprecated features as well.

We can communicate the change. As common sense, users should use upper bounds for all dependencies and consult changelogs. As this will be a big version increase, I think there should be no surprise if bounds are set.

Rebased & add warning.