Integrate OCRS backend and bundle Rust extension

[manykarim]

Summary

• add a Rust ocrs_py crate, Python adapter, and build hook so the packaged wheel ships the compiled OCRS engine and exposes it via DocTest.ocrs_adapter
• update DocumentRepresentation, VisualTest, configuration defaults, and Robot acceptance assets to target OCRS while keeping the legacy "tesseract" name as a deprecated alias
• refresh developer docs and pytest fixtures/tests to cover the new adapter, provide deterministic stubs when the engine/models are unavailable, and skip image-heavy suites when codecs or sample PDFs are missing

Testing

• poetry run pytest

---

https://chatgpt.com/codex/tasks/task_e_68dbb6e3aad48331bfab4f6cfb14f424