[exporter/datadog] Move configuration structs to main package

[mx-psi]

Description:

• Move configuration structs to main package of exporter. Deprecate config package.
• Remove deprecated Sanitize method by making it private (renamed to logWarnings).

Link to tracking Issue: Updates #8373

[codeboten]

Please rebase

[gbbr]

This is a good change, I think 🙂

[mx-psi]

@codeboten did you want to leave a review here?

[mx-psi]

@djaglowski This PR needs two release notes: one for deprecating the config package, and another for the removal of the Sanitize method. The two are closely related and are tracked by the same issue. What should I do?

[TylerHelmuth]

@djaglowski This PR needs two release notes: one for deprecating the config package, and another for the removal of the Sanitize method. The two are closely related and are tracked by the same issue. What should I do?

Related to this comment: #11847 (comment). For now I think you have to modify the changelog manually and then add a skip changelog label. Then then the release manager had to do some manual merging during the release.

[mx-psi]

Thanks @TylerHelmuth! I added Skip Changelog to bypass the changelog check, since that also sounds like a valid solution. I will wait for Daniel to see this before merging just in case :)

[djaglowski]

I agree w/ what Tyler said for now. Longer term, this is obviously not ideal.

Should we just remove the requirement that only 1 new entry is added? I think it's worth continuing to reject modifications/removals, but adding multiple change entries is probably allowable. I'm not sure we necessarily need to optimize for this case either, since it is infrequent enough. It seems like adding two files with some duplication between them is not unreasonable. WDYT?

[mx-psi]

Should we just remove the requirement that only 1 new entry is added? I think it's worth continuing to reject modifications/removals, but adding multiple change entries is probably allowable. I'm not sure we necessarily need to optimize for this case either, since it is infrequent enough. It seems like adding two files with some duplication between them is not unreasonable. WDYT?

I think that makes sense, yes, thanks for having a look! Modifications/removals are more likely to be mistakes (although there could be legitimate uses but they are very rare)