v1.7.0

# What's New in v1.7.0

This release brings unified batching across all ONNX models for efficient bulk processing, strengthened pipeline validation, improved classification accuracy, and enhanced transliteration capabilities. It also includes a comprehensive expansion of test coverage and developer tooling improvements.

## Highlights

- **API-Compatible Batching Across ONNX Models** – added a unified batching interface across embeddings, NER, POS tagging, dependency parsing, and transliteration models. Process multiple inputs in a single API call for significant performance improvements. All models now accept lists of inputs and return batched results consistently (e.g., `embedder.embed(['متن اول', 'متن دوم'])` returns embeddings for both).

- **Enhanced Pipeline Validation** – the `Pipeline` class now includes robust validation for step configuration, catching errors early with clear, actionable error messages. Improved error handling across all pipeline stages ensures better debugging and reliability in production workflows.

- **Improved Rule-Based Informal Classification** – the `RuleBasedInformalClassifier` now features enhanced accuracy for detecting informal and colloquial Persian text through improved keyword matching and better conjugated verb form detection.

- **Enhanced BYT5 Transliteration** – improvements to the `FarsiToTajik` and `TajikToFarsi` transliterators with better character-level encoding and decoding logic, resulting in more accurate transliteration of edge cases.

- **ALBERT Tokenizer Enhancements** – improved ALBERT tokenizer performance with better handling of special tokens and improved tokenization accuracy across various Persian text formats.

- **Comprehensive Test Suite Expansion** – added 400+ new test cases including:
  * 173 integration tests for batching contracts
  * 409 unit tests for batching validation
  * 51 pipeline-specific tests
  * Enhanced morphology and transliteration test coverage

## Other Changes

- Minimum Python version remains **3.11+**.
- Updated `ruff` to `0.16.1` for improved linting and code quality.
- Improved GitHub Actions workflow configuration for better CI/CD reliability.
- Expanded documentation and example notebooks covering new batching features.