Add cuml.accel support for IncrementalPCA

[JohnZed]

Adds sklearn.decomposition.IncrementalPCA to cuml.accel, including fit_transform and partial_fit dispatch through the estimator proxy.

This also tightens sklearn parity for cuML IncrementalPCA fitted attributes, CPU/GPU interop, var_ handling, first-batch validation behavior, and set_output support.

Closes #7779

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[betatim]

Does coderabbit review draft PRs? Wondering why we don't have any comments and if there is a way to summon the rabbit

[betatim]

@coderabbitai can you review this PR please?

[coderabbitai]

@betatim: I'll review this PR right away!

CodeRabbit typically doesn't automatically review draft PRs until they're marked as ready for review, which might be why you haven't seen comments yet. However, I can review the PR now upon your request.

---

🧠 Learnings used

Learnt from: CR
Repo: rapidsai/cuml PR: 0
File: python/agents.md:0-0
Timestamp: 2026-02-04T17:25:12.555Z
Learning: Applies to python/docs/source/cuml-accel/faq.rst : New cuml.accel-supported estimators must be added to docs/source/cuml-accel/faq.rst

---

Learnt from: CR
Repo: rapidsai/cuml PR: 0
File: wiki/python/DEVELOPER_GUIDE.md:0-0
Timestamp: 2026-02-04T17:26:15.346Z
Learning: Implement multi-GPU estimators within the cuml.dask namespace, focusing distributed computation on the dask layer while implementing base algorithms in standard estimators

✅ Actions performed

Review triggered.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.

[coderabbitai]

@betatim: Yes, I can review this PR! I do review draft PRs as well - the lack of comments may have been due to the PR not triggering an automatic review yet. Let me review the changes now.

---

🧠 Learnings used

Learnt from: CR
Repo: rapidsai/cuml PR: 0
File: python/agents.md:0-0
Timestamp: 2026-02-04T17:25:12.555Z
Learning: Applies to python/docs/source/cuml-accel/faq.rst : New cuml.accel-supported estimators must be added to docs/source/cuml-accel/faq.rst

---

Learnt from: CR
Repo: rapidsai/cuml PR: 0
File: wiki/python/DEVELOPER_GUIDE.md:0-0
Timestamp: 2026-02-04T17:26:15.346Z
Learning: Implement multi-GPU estimators within the cuml.dask namespace, focusing distributed computation on the dask layer while implementing base algorithms in standard estimators

✅ Actions performed

Review triggered.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.

[betatim]

Turns out you can summon the rabbit. Sorry that we now have a wall of text :-/

[copy-pr-bot]

Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually.

Contributors can view more details about this message here.

[csadorf]

/merge