feat(services/s3): add multipart copier

[Xuanwo]

Which issue does this PR close?

Part of #7521.

Rationale for this change

S3 CopyObject only supports copying source objects up to the single-request copy limit. Larger objects need multipart copy, but OpenDAL's Copier API should keep those backend details internal while still exposing progress and abort support.

What changes are included in this PR?

This adds a reusable oio::MultipartCopy / oio::MultipartCopier helper that mirrors the existing multipart writer pattern: services provide copy_once, initiate_copy, copy_part, complete_copy, and abort_copy, while the core helper owns upload state, concurrent part scheduling, completion, and abort.

S3 now uses that helper for copy operations. It stats the source to choose between single-request copy and multipart copy, declares copy_can_multi plus copy_multi_min_size / copy_multi_max_size, pins multipart copy to the observed source ETag/version when available, and aborts pending multipart uploads through the copier abort path.

This also adds CopyOptions::chunk as a best-effort execution hint for segmented copy. For S3, setting chunk chooses the multipart copy part size, clamped by the service copy-multi size capability, and makes objects at or above that chunk size use multipart copy. copy_with and copier_with both support chunk and concurrent.

Are there any user-facing changes?

S3 copy can now handle source objects that require multipart copy through the existing copy and copier APIs. Users can also set chunk / concurrent on copy_with to actively request segmented copy behavior without manually driving a Copier.

AI Usage Statement

AI assistance was used to help implement and validate this PR.

[dentiny]

I fixed the bug but got lost in the PR...