fix: extend Polly timeouts for local Ollama chat inference

qwen2.5-coder:7b consistently takes >30s, causing Polly's default
TotalRequestTimeout to reject every chat response. Override via
PostConfigureAll<HttpStandardResilienceOptions> when UseLocalAI=true
(dev-only path):
- TotalRequestTimeout: 30s  10min
- AttemptTimeout: 10s  5min
- CircuitBreaker.SamplingDuration: 30s  11min (Polly requires >= 2x AttemptTimeout)

The global override is acceptable here: this code path only runs
when the Ollama local-AI flag is set, which is developer-only.

Author: BenjaminMichaelis

EssentialCSharp.Web/Program.cs

@@ -19,6 +19,7 @@
 using Microsoft.AspNetCore.Diagnostics.HealthChecks;
 using Microsoft.EntityFrameworkCore;
 using Microsoft.Extensions.Diagnostics.HealthChecks;
+using Microsoft.Extensions.Http.Resilience;
 using OpenTelemetry;
 using OpenTelemetry.Instrumentation.AspNetCore;
 using OpenTelemetry.Metrics;
@@ -243,6 +244,21 @@ private static void Main(string[] args)
         // AIOptions__UseLocalAI=true enables Ollama local mode (set via aspire secret or dashboard).
         builder.AddAIServices(configuration);
 
+        // When using local Ollama, Polly's default 30s TotalRequestTimeout fires before LLM inference
+        // completes (qwen2.5-coder:7b consistently takes >30s). Override globally — this code path
+        // is only reached in local dev when UseLocalAI=true, so widening all clients is acceptable.
+        var aiOptsForTimeout = configuration.GetSection("AIOptions").Get<EssentialCSharp.Chat.AIOptions>();
+        if (aiOptsForTimeout?.UseLocalAI == true)
+        {
+            builder.Services.PostConfigureAll<HttpStandardResilienceOptions>(options =>
+            {
+                options.TotalRequestTimeout.Timeout = TimeSpan.FromMinutes(10);
+                options.AttemptTimeout.Timeout = TimeSpan.FromMinutes(5);
+                // Polly requires SamplingDuration >= 2x AttemptTimeout; default 30s is now invalid.
+                options.CircuitBreaker.SamplingDuration = TimeSpan.FromMinutes(11);
+            });
+        }
+
         // Add Rate Limiting for API endpoints
         builder.Services.AddRateLimiter(options =>
         {