fix(ci): default to all tests on all platforms + align with CLI naming

- Default test_type: quick → all (run full validation by default)
- Default os: linux → all (test both platforms by default)
- Rename 'release' → 'user-journey' to match CLI --user-journey flag
- Fix 'standard' to correctly stop at stage 5 (was running user-journey)
- Reorder options to show defaults first in dropdown

.github/workflows/tinytorch-validate-dev.yml

@@ -56,29 +56,29 @@ on:
       os:
         description: 'Operating system (linux, windows, all)'
         required: false
-        default: 'linux'
+        default: 'all'
         type: string
   workflow_dispatch:
     inputs:
       test_type:
-        description: 'Test type to run'
+        description: 'Test type to run (matches: tito dev test --<type>)'
         required: false
-        default: 'quick'
+        default: 'all'
         type: choice
         options:
-          - quick       # unit + cli (fast feedback, no build)
-          - standard    # stages 1-5 (no milestones)
-          - all         # stages 1-6 (everything except release)
-          - release     # destructive full validation
+          - all           # --all: Full validation (stages 1-7)
+          - quick         # --unit --cli: Fast feedback, no inline build
+          - standard      # Stages 1-5: inline + unit + integration + cli + e2e
+          - user-journey  # --user-journey: Full destructive journey test
       os:
         description: 'Operating system to test'
         required: false
-        default: 'linux'
+        default: 'all'
         type: choice
         options:
-          - linux       # Ubuntu only (default)
-          - windows     # Windows only (experimental)
-          - all         # Both Linux and Windows
+          - all         # Both Linux and Windows (default)
+          - linux       # Ubuntu only
+          - windows     # Windows only
 
 permissions:
   contents: read
@@ -161,7 +161,7 @@ jobs:
           esac
 
           # Set flags based on test type
-          # Sequential order: inline → unit → integration → cli → e2e → release
+          # Stages: inline → unit → integration → cli → e2e → fresh-install → user-journey
           case "$TEST_TYPE" in
             quick)
               # Fast feedback: skip inline build, just unit + cli
@@ -173,25 +173,16 @@ jobs:
               echo "run_release=false" >> $GITHUB_OUTPUT
               ;;
             standard)
-              # Stages 1-7 (includes user journey)
+              # Stages 1-5: inline + unit + integration + cli + e2e
               echo "run_inline=true" >> $GITHUB_OUTPUT
               echo "run_unit=true" >> $GITHUB_OUTPUT
               echo "run_integration=true" >> $GITHUB_OUTPUT
               echo "run_cli=true" >> $GITHUB_OUTPUT
               echo "run_e2e=true" >> $GITHUB_OUTPUT
-              echo "run_release=true" >> $GITHUB_OUTPUT
+              echo "run_release=false" >> $GITHUB_OUTPUT
               ;;
-            all)
-              # All stages 1-7 (used for dev/main pushes)
-              echo "run_inline=true" >> $GITHUB_OUTPUT
-              echo "run_unit=true" >> $GITHUB_OUTPUT
-              echo "run_integration=true" >> $GITHUB_OUTPUT
-              echo "run_cli=true" >> $GITHUB_OUTPUT
-              echo "run_e2e=true" >> $GITHUB_OUTPUT
-              echo "run_release=true" >> $GITHUB_OUTPUT
-              ;;
-            release)
-              # Full validation: all stages + destructive release at the end
+            all|user-journey)
+              # All stages 1-7 (full validation including destructive user journey)
               echo "run_inline=true" >> $GITHUB_OUTPUT
               echo "run_unit=true" >> $GITHUB_OUTPUT
               echo "run_integration=true" >> $GITHUB_OUTPUT