Card: FluidUse demos (topic sort, audio search, code search)

Add images: EmbeddingGemma2Vision (vision_70/140/280, GPU fp16) + position tables

Card: text + audio

Add audio: EmbeddingGemma2Audio (USM conformer, log-mel in-graph, GPU fp16, 617x real time)

EmbeddingGemma 2 text encoder on Core ML (Neural Engine): 7 shared-weight functions, bf16 token table