Extend image support to more components

[anakin87]

In Haystack 2.16.0, the foundations of image support have been integrated. #8976

We want to bring image support to more components, namely Chat Generators and Embedders.

Notes

• I already verified that image support works with AzureOpenAIChatGenerator (since inherits from OpenAIChatGenerator)
• MetaLlamaChatGenerator already works
• MistralChatGenerator works when used with a multimodal model
• NvidiaChatGenerator already works
• OpenRouterChatGenerator already works (tested with Llama 3.2 11B Vision Instruct and Qwen2.5 VL 72B Instruct)
• STACKITChatGenerator already works (even if they only offer google/gemma-3-27b-it)