fix(regression): Make test_analyze_mv row_count assertion stable after truncate

[yujun777]

Problem

statistics/test_analyze_mv.groovy line 614 asserts assertEquals("-1", result_row[0][4]) immediately after truncate table, expecting report_row_count_for_nereids to be -1 (unreported). This is a race condition: BE asynchronously reports new tablet stats (0 rows for empty table) to FE, and if the report arrives before the assertion, the value is 0 instead of -1.

On cloud, this is amplified by CloudTabletStatMgr unconditionally setting rowCountReported=true, making the -1 state exceptionally short-lived or unobservable.

Root Cause

After truncate table, the FE stat manager (CloudTabletStatMgr / TabletStatMgr) updates MaterializedIndex.rowCount and sets rowCountReported=true. The test assertion races with this update:

• If BE hasn't reported yet → getRowCountForIndex(id, true) returns -1 ✓
• If BE has reported → returns 0 ✗ (test fails)

Fix

Change the assertion to accept both -1 and 0, since both are valid states for an empty table after truncate:

• -1: tablet row count not yet reported
• 0: tablet row count reported as 0 (empty table)

Add assertion message with actual value for debuggability.

🤖 Generated with Claude Code

[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?

[yujun777]

run buildall

[yujun777]

run nonConcurrent

[morrySnow]

/review

[github-actions]

PR approved by at least one committer and no changes requested.

[github-actions]

PR approved by anyone and no changes requested.

[github-actions]

Code review completed. I found no blocking issues in this PR.

Critical checkpoint conclusions:

• Goal and proof: The change addresses a real flaky assertion after truncate table; show index stats strict row-count can be -1 before tablet row-count reporting completes or 0 after the empty-table report arrives. The suite still later waits for and asserts the stable 0 state for both the MV and base index.
• Scope: The modification is small and focused, limited to the flaky regression assertion.
• Concurrency/lifecycle: The relevant behavior is asynchronous tablet-stat reporting after truncate. No new production concurrency or lifecycle code is introduced.
• Config/compatibility/protocol: No configuration, storage format, protocol, or FE/BE variable-passing changes.
• Parallel paths: The later stable row-count checks remain unchanged, so this only relaxes the immediate race window and does not remove end-state coverage.
• Test coverage/results: This is a regression-test-only fix; no .out result update is needed for this assertion-only change.
• Observability/performance/transaction/persistence: No new production behavior is introduced, so these are not materially affected.

User focus points: no additional user-provided review focus was present.