perf: Increase benchmark iterations to 100 for statistical significance

- FirstCatalogQuery: 100 fresh processes (was 10)
- SubsequentCatalogQuery: 100 iterations (was 20)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: haoliuu

.github/workflows/foundry-local-benchmark.yml

@@ -63,30 +63,30 @@ jobs:
     - name: Build Test Project
       run: dotnet build AIDevGallery.Tests -r win-x64 -f net9.0-windows10.0.26100.0 /p:Configuration=Release /p:Platform=x64
 
-    - name: Run FirstCatalogQuery (10 fresh processes)
+    - name: Run FirstCatalogQuery (100 fresh processes)
       shell: pwsh
       run: |
         $env:PERFORMANCE_OUTPUT_PATH = "${{ github.workspace }}\PerfResults"
         New-Item -ItemType Directory -Force -Path $env:PERFORMANCE_OUTPUT_PATH | Out-Null
 
         $testDll = "AIDevGallery.Tests\bin\x64\Release\net9.0-windows10.0.26100.0\win-x64\AIDevGallery.Tests.dll"
 
-        for ($i = 1; $i -le 10; $i++) {
-          Write-Host "`n=== FirstCatalogQuery Run $i/10 ==="
+        for ($i = 1; $i -le 100; $i++) {
+          Write-Host "`n=== FirstCatalogQuery Run $i/100 ==="
           dotnet test $testDll `
             --filter "TestCategory=FirstCatalogQuery" `
             --logger "console;verbosity=detailed" `
             --verbosity quiet
           Write-Host "Exit code: $LASTEXITCODE"
         }
 
-    - name: Run SubsequentCatalogQuery (20 iterations in one process)
+    - name: Run SubsequentCatalogQuery (100 iterations in one process)
       shell: pwsh
       run: |
         $env:PERFORMANCE_OUTPUT_PATH = "${{ github.workspace }}\PerfResults"
         $testDll = "AIDevGallery.Tests\bin\x64\Release\net9.0-windows10.0.26100.0\win-x64\AIDevGallery.Tests.dll"
 
-        Write-Host "`n=== SubsequentCatalogQuery (20 iterations) ==="
+        Write-Host "`n=== SubsequentCatalogQuery (100 iterations) ==="
         New-Item -ItemType Directory -Force -Path "${{ github.workspace }}\TestResults" | Out-Null
         dotnet test $testDll `
           --filter "TestCategory=SubsequentCatalogQuery" `

AIDevGallery.Tests/IntegrationTests/FoundryLocalPerformanceBenchmark.cs

@@ -106,7 +106,7 @@ public async Task BenchmarkFirstCatalogQuery()
 
     /// <summary>
     /// Measures subsequent catalog queries after initialization (cached/warm path).
-    /// Runs 20 iterations in a single process.
+    /// Runs 100 iterations in a single process.
     /// </summary>
     /// <returns>A <see cref="Task"/> representing the asynchronous operation.</returns>
     [TestMethod]
@@ -142,7 +142,7 @@ public async Task BenchmarkSubsequentCatalogQuery()
         // Warm-up call (not measured)
         await catalog.ListModelsAsync();
 
-        for (int i = 0; i < 20; i++)
+        for (int i = 0; i < 100; i++)
         {
             var sw = Stopwatch.StartNew();
             var models = await catalog.ListModelsAsync();