Text this: OSS–CAEA: Bridging Vision and Language for Open‐Vocabulary Semantic Segmentation via Collaborative Attention and Embedding Alignment