Add new components (TOC Refiner, Sentence and Recursive chunker, cohera and Voyage Embedders)!)

Added

• SentenceChunker: Sentence-aware sliding-window chunker with configurable sentences_per_chunk, sentence_overlap, and min_chunk_size. Pure-Python regex splitting, no external dependencies.
• RecursiveMarkdownChunker: Hierarchical chunker that splits by heading level (H1→H2→H3→H4), then paragraph, then sentence, with overlap merging.
• VoyageAIEmbedder: Embedder using Voyage AI models — voyage-3, voyage-3-large, voyage-3-lite.
• CohereEmbedder: Embedder using Cohere models — embed-v4.0, embed-english-v3.0, embed-multilingual-v3.0. Supports configurable input_type for search documents vs queries.
• TableOfContentsRefiner: Refiner that detects and removes the table of contents section, storing it in extracted_data["toc_markdown"].
• Example scripts (examples/): 4 runnable Python scripts demonstrating basic pipeline usage, component comparison, step-by-step execution, and cost tracking.
• Jupyter notebooks (notebooks/): 3 interactive notebooks — getting started walkthrough, chunker comparison, and a component explorer exercising every component with all valid configurations.
• README component reference tables: OCR, Refiners, Chunkers, and Embedders with models, parameters, defaults, and costs per 1M tokens.
• README examples & notebooks section linking to all new files with descriptions.

Changed

• Updated README package layout tree to include examples/, notebooks/, and expanded src/ragbandit/ subdirectories

Dependencies

• Added voyageai for Voyage AI embeddings support
• Added cohere for Cohere embeddings support