XGBoost integration infrastructure (still temporarily using Estimator rather than Trainer)

[rgesteve]

We are excited to review your PR.

So we can do the best job, please check:

• There's a descriptive title that will make sense to other developers some time from now.
• [] There's associated issues. All PR's should have issue(s) associated - unless a trivial self-evident change such as fixing a typo. You can use the format Fixes #nnnn in your description to cause GitHub to automatically close the issue(s) when your PR is merged.
• Your change description explains what the change does, why you chose your approach, and anything else that reviewers should know.
• You have included any necessary tests in the same PR.

Strictly this is independent of OneDAL (XGBoost and OneDAL are separate libraries), but adding to this feature branch for convenience. This payload includes the PInvoke infrastructure to call XGBoost (version 1.5 and above... current version is 1.7). It mimics the design of the LightGBM integration as much as possible. Most notable difference is that it does not serialize to InternalEnsembleTree, but uses its own model (also it uses the JSON dump to build such a model, but that's an implementation detail). The actual Estimator/Transformer pair is preliminary, as it looks more like a standard transformation than a learner (in that the Model is not an independent entity). This most likely will need to change.

[LittleLittleCloud]

Do you want to merge it to onedal-integration or main branch?

[rgesteve]

To onedal-integration, please...

[rgesteve]

Closing in favor of #6383, #6364 and #6373 which include only the relevant files to the changes (rather than the noise introduced by merging from feature branch)