chore: bump to 1.4.6

Release highlights since 1.4.5:
- Transcript: OpenRouter and Google model fetchers now store token
  limits even when the API omits a max-output figure, so the model
  parameters panel prefills real values.
- Settings: removed the duplicate standalone "Max tokens" field; the
  "Model parameters" panel is the single token-limit control on every
  provider.
- Model/provider notices reworded to "output budget" so the figure is
  honestly labelled as a derived value, not a hidden setting.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

manifest.json

@@ -1,7 +1,7 @@
 {
     "id": "tubesage",
     "name": "TubeSage",
-    "version": "1.4.5",
+    "version": "1.4.6",
     "minAppVersion": "1.11.4",
     "description": "Create comprehensive notes from YouTube transcripts using LLMs.",
     "author": "Richard McCorkle",

package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "tubesage",
-  "version": "1.4.5",
+  "version": "1.4.6",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "tubesage",
-      "version": "1.4.5",
+      "version": "1.4.6",
       "license": "MIT",
       "dependencies": {
         "@langchain/core": "^0.3.17",

package.json

@@ -1,6 +1,6 @@
 {
   "name": "tubesage",
-  "version": "1.4.5",
+  "version": "1.4.6",
   "description": "Create comprehensive notes from YouTube transcripts using LLMs.",
   "main": "main.js",
   "scripts": {