Allow individual entities to be excluded from llms.txt Data Types section

Title: Currently, the llms.txt file auto-includes all entities defined in the app with no exclusion mechanism. For apps with significant internal infrastructure, this exposes operational entities that have no value to AI crawlers and misrepresent what the app does to users. In our case, 67 entities are listed including CronStatus, RoutingLog, BackfillState, ProfileUpdateQueue, RadarScoringJob, DataCurationStatus, EntitlementAuditLog, and others that are purely operational. When AI engines like ChatGPT, Gemini, and Perplexity read this file to understand what our app does, they're reading infrastructure noise alongside the data types that actually matter to users. Requested: Add an exclude_from_llms boolean field (default false) to the entity schema configuration. When set to true, the entity is omitted from the llms.txt Data Types section. No other behavior changes. This is a small schema-level flag that would meaningfully improve AI discoverability for any Base44 app with operational infrastructure.