Preserve FTS restoration failure diagnostics

[Widthdom]

Summary

• preserve the first actionable full-index exception when the implicit FTS cleanup retry also fails
• validate newly persisted failure diagnostics and fall back to a bounded stackless exception chain when a platform frame is not report-safe
• add regression coverage, bilingual user/testing documentation, and changelog.d/unreleased/5012.fixed.md

Root cause

A failure from FtsBulkLoadTriggerGuard.Complete left the guard active so the surrounding using retried cleanup. If that retry also failed, Dispose replaced the original exception. Separately, the failure writer could persist a sanitized diagnostic shape that the report reader's stricter canonical validation rejected as invalid.

Validation

• dotnet build CodeIndex.sln -c Release --no-restore -p:UseSharedCompilation=false
• dotnet build CodeIndex.sln -c Debug --no-restore -p:UseSharedCompilation=false
• new Issue Full index can fail while restoring FTS triggers after graph finalization #5012 tests passed on net8.0 and net9.0
• Codex adversarial review reran the related database/report suites: 364/364 passed on each target
• full Release suite completed with only pre-existing Issue Avoid overhead that makes parallel batch slower than sequential #4872 parallel-wave timeouts under concurrent-suite load; both affected tests passed 2/2 when rerun alone on each target
• dotnet format CodeIndex.sln --verify-no-changes --no-restore
• dotnet run --project tools/CodeIndex.Changelog -- check
• two large existing-database index updates completed graph finalization and FTS restoration with zero warnings/errors (20m24s and 16m10s)
• Codex adversarial review: No blocking/actionable issues found.

Fixes #5012