feat(canvas): add video-prompt node and enhance video generation support

- Introduced a new node type "video-prompt" for AI video generation, including its integration into the canvas command palette and node template picker.
- Updated connection validation to allow connections from text nodes to video-prompt and from video-prompt to ai-video nodes.
- Enhanced error handling and messaging for video generation failures, including specific cases for provider issues.
- Added tests to validate new video-prompt functionality and connection policies.
- Updated localization files to include new labels and prompts for video-prompt and ai-video nodes.

convex/schema.ts

@@ -222,6 +222,12 @@ export default defineSchema({
     canvasId: v.optional(v.id("canvases")),        // Zugehöriger Canvas
     openRouterCost: v.optional(v.number()),        // Tatsächliche API-Kosten (Cent)
     model: v.optional(v.string()),                 // OpenRouter Model ID
+    provider: v.optional(v.union(v.literal("openrouter"), v.literal("freepik"))),
+    videoMeta: v.optional(v.object({
+      model: v.string(),
+      durationSeconds: v.number(),
+      hasAudio: v.boolean(),
+    })),
   })
     .index("by_user", ["userId"])
     .index("by_user_type", ["userId", "type"])