Default to extract + clean + watch

Without clean, the code diff is really ugly and hard to track

README.md

@@ -100,13 +100,12 @@ Everything is designed and engineered following my taste and vision. This is a p
 Prerequisites: Node.js 18+
 
 - `npm install` - Install dependencies
-- `npm run dev` - Start development server and `messages:extract:watch` in parallel
+- `npm run dev` - Start development server and `messages:extract` (`clean` + ``watch`) in parallel
 - `npm run build` - Build for production
 - `npm run preview` - Preview the production build
 - `npm run fetch-instances` - Fetch instances list from [joinmastodon.org/servers](https://joinmastodon.org/servers), save it to `src/data/instances.json`
 - `npm run sourcemap` - Run `source-map-explorer` on the production build
 - `npm run messages:extract` - Extract messages from source files and update the locale message catalogs
-- `npm run messages:extract:watch` - Same as `messages:extract` but in watch mode
 
 ## Tech stack
 

package.json

@@ -4,14 +4,13 @@
   "type": "module",
   "scripts": {
     "dev:vite": "vite",
-    "dev": "run-p dev:vite messages:extract:watch",
+    "dev": "run-p dev:vite \"messages:extract:clean -- --watch\"",
     "build": "vite build",
     "preview": "vite preview",
     "fetch-instances": "env $(cat .env.local | grep -v \"#\" | xargs) node scripts/fetch-instances-list.js",
     "sourcemap": "npx source-map-explorer dist/assets/*.js",
     "bundle-visualizer": "npx vite-bundle-visualizer",
     "messages:extract": "lingui extract",
-    "messages:extract:watch": "lingui extract --watch",
     "messages:extract:clean": "lingui extract --clean",
     "messages:compile": "lingui compile"
   },