Feature Request: add support for vision model InternVL3_5

[kiron111]

Prerequisites

• I am running the latest code. Mention the version if possible as well.
• I carefully followed the README.md.
• I searched using keywords relevant to my issue to make sure that I am creating a new issue that is not already open (or closed).
• I reviewed the Discussions, and have a new and useful enhancement to share.

Feature Description

Some of them are MOE (some are based on Qwen3):
https://huggingface.co/OpenGVLab/InternVL3_5-30B-A3B
https://huggingface.co/OpenGVLab/InternVL3_5-241B-A28B

It should be now SOTA in open source VL models (based on InternLM Lab's report)

GGUF here:
https://huggingface.co/QuantStack/InternVL3_5-30B-A3B-gguf/tree/main

It would be great if ik_llama.cpp would support it.
Thankyou!!

Motivation

it could be run in mainline llama.cpp (tesed) , but the speed is not optimized, could be greatly enhanced in ik-llama.cpp in hybrid CPU+GPU inference.

Possible Implementation

No response

[ikawrakow]

This can be addressed after merging #798.

[kiron111]

This can be addressed after merging #798.

thankyou!!