fix(audit): swap Perplexity sonar-pro → sonar-reasoning-pro

sonar-pro's 8K output cap can't fit the full AEO audit JSON. Switch
to sonar-reasoning-pro, which has a higher output capacity and is
the right tier for a structured analytical task. The 32K max_tokens
from the previous commit now has actual headroom to use.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: ralyodio

lib/audit/perplexity-engine.ts

@@ -9,7 +9,7 @@ export async function perplexityAudit(targetUrl: string): Promise<ClaudeAuditRes
   return oaCompatAudit(targetUrl, {
     apiKey: env.perplexityApiKey,
     baseURL: "https://api.perplexity.ai",
-    model: "sonar-pro",
+    model: "sonar-reasoning-pro",
     providerLabel: "Perplexity",
     maxOutputTokens: 32_000,
     responseFormat: aeoAuditResponseFormat(),