OpenVINO GenAI: Add Image-to-Video Support to LTX Video Generation Pipeline

OpenVINO GenAI currently supports text-to-video generation via the LTX pipeline but lacks image conditioning. This project adds Image-to-Video (I2V) generation, enabling users to produce videos from an input image combined with a text prompt, running on Intel CPU and GPU. The implementation follows the HuggingFace Diffusers LTXImageToVideoPipeline reference: the input image is VAE-encoded once, and the resulting latent is prepended as frame 0 along the frame axis of the packed latent tensor throughout the denoising loop. Deliverables include: VAE encoder support in AutoencoderKLLTXVideo, a generate(image, prompt, config) overload in both C++ and Python APIs, strength-based timestep scheduling, C++ and Python samples, WWB and LLM benchmarking updates, and a cross-language test suite validating numerical consistency between C++ and Python implementations.