fix: replace implicit app inference with explicit selection for Skills import and sync

Skills import previously inferred app enablement from filesystem presence,
causing incorrect multi-app activation when the same skill directory existed
under multiple app paths. Now the frontend submits explicit app selections
via ImportSkillSelection, and schema migration preserves a snapshot of
legacy app mappings to avoid lossy reconstruction.

Also adds reconciliation to sync_to_app (removes disabled/orphaned symlinks)
and MCP sync_all_enabled (removes disabled servers from live config).

src-tauri/src/commands/skill.rs

@@ -6,7 +6,9 @@
 
 use crate::app_config::{AppType, InstalledSkill, UnmanagedSkill};
 use crate::error::format_skill_error;
-use crate::services::skill::{DiscoverableSkill, Skill, SkillRepo, SkillService};
+use crate::services::skill::{
+    DiscoverableSkill, ImportSkillSelection, Skill, SkillRepo, SkillService,
+};
 use crate::store::AppState;
 use std::sync::Arc;
 use tauri::State;
@@ -85,10 +87,10 @@ pub fn scan_unmanaged_skills(
 /// 从应用目录导入 Skills
 #[tauri::command]
 pub fn import_skills_from_apps(
-    directories: Vec<String>,
+    imports: Vec<ImportSkillSelection>,
     app_state: State<'_, AppState>,
 ) -> Result<Vec<InstalledSkill>, String> {
-    SkillService::import_from_apps(&app_state.db, directories).map_err(|e| e.to_string())
+    SkillService::import_from_apps(&app_state.db, imports).map_err(|e| e.to_string())
 }
 
 // ========== 发现功能命令 ==========