Drop the deprecated NoneShardSpec

[jihoonson]

A part of #9241. NoneShardSpec has been deprecated for a while.

---

This PR has:

• been self-reviewed.
  • using the concurrency checklist (Remove this item if the PR doesn't have any relation to concurrency.)
• added documentation for new or modified features or behaviors.
• added Javadocs for most classes and all non-trivial methods. Linked related entities via Javadoc links.
• added or updated version, license, or notice information in licenses.yaml
• added comments explaining the "why" and the intent of the code wherever would not be obvious for an unfamiliar reader.
• added unit tests or modified existing tests to cover new code paths.
• added integration tests.
• been tested in a test Druid cluster.

[drcrallen]

This breaks compatibility with segments that were built with none shard spec. My understanding is that very old segments are still to be supported to load on historicals for query.

[jihoonson]

Yes, this is an incompatible change. The NoneShardSpec was deprecated in March last year (#6883). I don't think we should support everything forever. For example, we don't support the old segment format (lower than the version 8) anymore.
However, I'm open to keep it for a couple of more releases. Do you think the next release is not a good time to remove it?

[gianm]

I don't think we should ever remove support for loading any previously-existing segments, no matter how old they are. To my knowledge, we haven't done this since Druid was open-sourced (if I recall correctly, the pre-v8 segment versions were also pre-OSS). This is because it is a pretty big burden on users to ask them to reindex all their data.

Does the fact that NoneShardSpec exists block #9241 from working?

If so, an alternative would be to remove NoneShardSpec, but include code to automatically convert metadata into something equivalent, so users don't need to reindex.

[jihoonson]

Ok. This is not a blocker. I'm closing this PR.