CloudQueryError prevents /chronicle standup despite local fallback data

### Describe the bug

## Summary

`/chronicle standup` fails when the cloud session store returns `CloudQueryError: internal server error`.

Local session data is available, and some `session_store_sql` queries successfully return local fallback rows. However, the DuckDB-style timestamp predicates used for last-24-hours chronicle queries appear to prevent local fallback from rescuing the command.

## Environment

- GitHub Copilot CLI: `1.0.64`
- Platform: Windows
- Package metadata: `@github/copilot` `1.0.64`, build metadata `ae586e7`
- Startup log showed: `No update needed, current version is 1.0.64, fetched latest release is v1.0.64`

## Impact

`/chronicle standup` cannot generate a standup report even though recent local session rows are available.

## Expected behavior

When the cloud session store fails with an internal server error, `session_store_sql` and `/chronicle standup` should degrade gracefully to local indexed sessions where possible, returning local rows with a warning.

## Actual behavior

The required last-24-hours query fails with:

```text
CloudQueryError: {"documentation_url":"https://docs.github.com/rest","message":"internal server error"}
```

### Affected version

1.0.64

### Steps to reproduce the behavior

This query fails completely:

SELECT id, branch, updated_at
FROM sessions
WHERE updated_at >= TIMESTAMP '2026-06-23T07:47:34.106Z'
ORDER BY updated_at DESC
LIMIT 5

This also failed:

SELECT id, branch, updated_at
FROM sessions
WHERE updated_at >= now() - INTERVAL '1 day'
ORDER BY updated_at DESC
LIMIT 5

But this query returns local fallback rows:

SELECT id, branch, updated_at
FROM sessions
ORDER BY updated_at DESC
LIMIT 5

Observed fallback behavior:

5 row(s) returned from _query_source = local
Cloud query failed (CloudQueryError: {"documentation_url":"https://docs.github.com/rest","message":"internal server error"}). Showing local results only.

### Expected behavior

_No response_

### Additional context

_No response_

Commencez par la gestion de session_store_sql et le chemin de requête de /chronicle standup. Reproduisez les requêtes filtrées par horodatage et comparez-les à la requête de secours locale non filtrée ; c’est terminé lorsque /chronicle standup renvoie les sessions indexées localement avec un avertissement en cas d’échec de la requête cloud.

Rédigé par le modèle d'indexation à partir du texte de l'issue.