docs: Update INFERENCE data event reference for LiteRT & LiteRT-LM

[javoweb-fc]

Description

This PR updates the developer reference documentation for the INFERENCE data event function to align with the latest on-device AI runtime capabilities and contracts in the fulcrum-expressions and KMP models repositories.

The updated documentation has been streamlined to keep underlying backend engines (LiteRT/LiteRT-LM) completely transparent to developers, while providing a clear and user-friendly guide focusing strictly on Form Reference Files for model loading.

Key Changes

1. User-Friendly Abstraction:
   • Removed all direct references to LiteRT and LiteRT-LM in user-facing texts, replacing them with generic concepts like Vision ML and Generative LLM.
2. Strict File-Extension-Based Auto-Detection:
   • Documented that model type auto-detection is determined strictly by the file extension (.tflite for Vision ML vs .gguf/.litertlm/.task for Generative LLM).
   • Clarified that config-level parameters do not affect or override detection, but must still match the model type.
3. Streamlined Model Loading (Form Reference Files Only):
   • Removed any references to the central Model Catalog, catalog IDs, and resolution priority. The documentation now focuses exclusively on loading custom models via Form Reference Files.
4. ONNX Support Deprecation:
   • Marked the legacy/ONNX-based configurations as Deprecated and added guidance to update to the newer config-based model definitions.
5. Execution Modes:
   • Clearly separated the documentation into three clear paths: Vision ML, Generative LLM, and Legacy Vision ML (Deprecated).
6. Performance and Device Warnings:
   • Added a prominent, highly visible callout warning about computational overhead, device memory footprint, and battery drain.
7. Updated Examples:
   • Updated the standard Vision ML example to use the modern config syntax.
   • Added a new example demonstrating how to run an on-device LLM text summarization workflow (gemma-4-e2b.litertlm).

JIRA / Issue Reference

Closes: FLCRM-20632

Testing & Verification

• Verified alignment of all properties, data types, error cases, and structures with functions.coffee (fulcrum-expressions) and InferenceModelResolver.kt (kmp/models).
• Ran validation scans to ensure zero remaining references to "LiteRT", "LiteRT-LM", "Model Catalog", or "catalog ID" exist in user-facing text.

[Copilot]

Pull request overview

This PR updates the INFERENCE data event developer reference to reflect the current on-device inference runtimes (LiteRT for vision and LiteRT-LM for LLM-style text generation), adds clearer mode-specific parameter documentation, and refreshes examples accordingly.

Changes:

• Reworked the INFERENCE overview to distinguish Vision ML vs Generative LLM execution modes and added a prominent device resource/battery warning.
• Documented modern options.config-based parameter schemas for LiteRT (vision) and LiteRT-LM (LLM), and marked legacy ONNX usage as deprecated.
• Updated and expanded examples to show modern vision config usage and an on-device text summarization workflow.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated 5 comments.

[Copilot]

Pull request overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated 6 comments.

[Copilot]

Pull request overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated 3 comments.

[Copilot]

Pull request overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated 3 comments.

[Copilot]

Pull request overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated 4 comments.