Refactor ChunkingSettings classes from Inference plugin to XPack core

[kderusso]

This will allow chunking settings to be easily re-used by the ES|QL plugin.

[kderusso]

In order to make this PR as small as possible, I surgically moved only those methods necessary from ServiceUtils and kept callers there for actual implementation messages. I did not move tests either. If I move things more aggressively this PR gets exponentially bigger.

[elasticsearchmachine]

Pinging @elastic/ml-core (Team:ML)

[elasticsearchmachine]

Pinging @elastic/search-relevance (Team:Search - Relevance)

[DonalEvans]

Are there any follow-up PRs planned to move tests to the appropriate module and maybe move the rest of the ServiceUtils methods too?

While moving classes can create a PR that touches a lot of files, the actual changes themselves are very easy to review, so I wouldn't mind including them as part of this PR if that would make it easier to keep everything where it should be.

[kderusso]

@DonalEvans Sure, I can move the tests too. I didn't at first because there are still a lot of callers in ServiceUtils but agree it's cleaner.

[DonalEvans]

Would it be possible to also move the tests for the various chunking settings classes that were moved in this PR? Specifically to move these test classes from xpack/inference/chunking to xpack/core/inference/chunking:

• ChunkingSettingsBuilderTests
• NoneChunkingSettingsTests
• RecursiveChunkingSettingsTests
• SentenceBoundaryChunkingSettingsTests
• WordBoundaryChunkingSettingsTests