[fix](load) Respect Kafka isolation level when fetching offsets

[sollhui]

What problem does this PR solve?

Issue Number: None

Related PR: None

Problem Summary:

Routine load used query_watermark_offsets() when resolving OFFSET_END and refreshing latest offsets. The high watermark is the log end offset and includes records from open Kafka transactions, so isolation.level=read_committed did not affect routine load progress, lag, or scheduling decisions.

This change uses Kafka’s ListOffsets Admin API and passes the job isolation level explicitly. read_committed now returns the last stable offset, while read_uncommitted continues to return the log end offset. Both periodic latest-offset refresh and OFFSET_END resolution share the same implementation.

A transactional Kafka regression case covers the different initial offsets and lag values for read_committed and read_uncommitted jobs.

Release note

Kafka routine load offset queries now respect property.isolation.level.

Check List (For Author)

• Test: Not run (skipped at requester direction)
• Behavior changed: Yes. read_committed offset and lag exclude records in open transactions.
• 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?