branch-4.1: [perf](load) Optimize write performance with improved backpressure control

[sollhui]

What problem does this PR solve?

Backport #66847 to branch-4.1.

For S3-backed cloud loads, a memtable flush can finish after submitting asynchronous uploads, so flush-running count does not represent downstream upload pressure. This change detects S3 storage and uses the S3 upload queue for backpressure, falling back to the existing flush-running limit when the upload pool is unavailable. HDFS retains flush-count backpressure.

Branch-4.1 adaptation

branch-4.1 does not contain WriteRequestType::GROUP or CloudGroupRowsetBuilder. The row-binlog GROUP-specific changes from the original PR are therefore omitted. CloudRowsetBuilder is final in this branch, so is_s3_storage is non-virtual. Local-storage flush limits remain unchanged.

Release note

Improve load backpressure for S3-backed cloud loads. Higher concurrency can increase memory use.

Check List (For Author)

• Test: Compilation and tests skipped as requested. clang-format 16 and git diff --check passed.
• Behavior changed: Yes; S3-backed loads use upload queue pressure. HDFS and local-storage limits remain unchanged.
• Does this need documentation: No.

[hello-stephen]

Thank you for your contribution to Apache Doris.
Don't know what should be done next? See How to process your PR.

Please clearly describe your PR:

1. What problem was fixed (it's best to include specific error reporting information). How it was fixed.
2. Which behaviors were modified. What was the previous behavior, what is it now, why was it modified, and what possible impacts might there be.
3. What features were added. Why was this function added?
4. Which code was refactored and why was this part of the code refactored?
5. Which functions were optimized and what is the difference before and after the optimization?

[sollhui]

run buildall