Restore-DbaDatabase - Honor EnableException in the backup information pipeline

[andreasjordan]

Summary

Fixes #10621. When Restore-DbaDatabase is given backups without a full to anchor the chain, Select-DbaBackupInformation correctly calls Stop-Function - but the splat Restore-DbaDatabase builds for it never passed EnableException on, so under -EnableException the stop degraded to a warning, control returned, and the restore proceeded over an empty $FilteredBackupHistory to a successful-looking no-op. A caller relying on the promised exception (a test fixture, or any script that restores and then uses the database) continued as if the restore had happened - exactly how Get-DbaDbRestoreHistory.Tests.ps1 was flaky for years (#10620).

What changed

• EnableException = $EnableException added to the Select-DbaBackupInformation splat and the Get-DbaBackupInformation splat, and to the per-file Get-DbaBackupInformation call on the SQL Server 2000 path (now a splat, since it grew to three parameters).
• The "No backups passed through" bail-out is now a Stop-Function instead of a plain Write-Message -Level Warning, so it too honors -EnableException.

What deliberately did not change

• Without -EnableException nothing changes: Stop-Function writes the identical warning and returns, so the default warn-and-return behavior is untouched. The open behavior question from Restore-DbaDatabase -EnableException restores nothing with only a warning when no full backup anchors the chain #10621 - whether an empty selection should also stop the non-exception path - is left for a maintainer decision.
• The Test-DbaBackupInformation call already passed EnableException = $true inside try/catch and stays as it is.
• Format-DbaBackupInformation contains no Stop-Function at all, so there is nothing to propagate.
• Get-DbaAgBackupHistory has the same gap on its Select-DbaBackupInformation call and gets its own pull request, one command per PR.

Tests

Three regression tests in a new context, each backup written to its own file so the minute-resolution default name cannot hide the missing full by merging it into the diff's file:

• diff+log without the full throws *Fullname property not found* with -EnableException (the exact repro from the issue),
• the same input still warns and returns nothing without it,
• an existing directory with no backups throws *No backups passed through* with -EnableException.

Full file verified via the lab harness on SQL03\SQL2019: 84 tests, 76 passed, 0 failed, 7 skipped (the documented StopAt and Azure skips), lab left clean.

created by Claude and reviewed by Andreas Jordan

🤖 Generated with Claude Code

[potatoqualitee]

Reviewed the complete current head, all backup-information propagation paths, empty-history behavior, regression coverage, and exact-head SQL CI evidence. No material correctness or compatibility issues found.