backend: fix a crash on inputs greater than n_ctx

[cebtenzzre]

This fixes a regression in commit 4fc4d94 ("fix chat-style prompt templates (#1970)"), which moved some return statements into a new function (LLModel::decodePrompt) without making them return from the parent as well.

---

🚀	This description was created by Ellipsis for commit 3409b98

Summary:

Fixes crash on inputs greater than n_ctx by updating LLModel::decodePrompt to return a boolean and handle errors appropriately.

Key points:

• Modified gpt4all-backend/llmodel.h:
  • Changed LLModel::decodePrompt return type from void to bool.
• Updated gpt4all-backend/llmodel_shared.cpp:
  • Updated LLModel::prompt to handle decodePrompt return value and exit on error.
  • Modified LLModel::decodePrompt to return false on errors and true on success.

---

Generated with ❤️ by ellipsis.dev