The bug was using size (remaining bytes) instead of length (actual part size) when constructing PartInfo in the
multipart copy loop. This would record wrong sizes for each part - especially problematic for the last part.
* error refactored: moved to thiserror
* Result type alias for better ergonomics:
* removed field from MinioErrorCode::BucketNotEmpty enum
* made field private of MinioErrorResponse
* updated XmlError
* simplified calling errors
* bumped toolchain channel form 1.86.0 to 1.87.0
* bumped toolchain channel form 1.87.0 to 1.88.0 due to clippy fixes that are not compatible with 1.87.0
