fix(database): clear provider health record when removing from failover queue

When a provider is removed from the failover queue, its health monitoring
is no longer needed. This change ensures the health record is also deleted
from the database to prevent stale data.

src-tauri/src/database/dao/failover.rs

@@ -78,12 +78,22 @@ impl Database {
     ) -> Result<(), AppError> {
         let conn = lock_conn!(self.conn);
 
+        // 1. 从队列中移除
         conn.execute(
             "UPDATE providers SET in_failover_queue = 0 WHERE id = ?1 AND app_type = ?2",
             rusqlite::params![provider_id, app_type],
         )
         .map_err(|e| AppError::Database(e.to_string()))?;
 
+        // 2. 清除该供应商的健康状态（退出队列后不再需要健康监控）
+        conn.execute(
+            "DELETE FROM provider_health WHERE provider_id = ?1 AND app_type = ?2",
+            rusqlite::params![provider_id, app_type],
+        )
+        .map_err(|e| AppError::Database(e.to_string()))?;
+
+        log::info!("已从故障转移队列移除供应商 {provider_id} ({app_type}), 并清除其健康状态");
+
         Ok(())
     }