Fix linting and type checking errors - modernize Python 2 compatibility code

[dmort27]

• What kind of change does this PR introduce? Code quality improvement, bug fix, modernization

• What is the current behavior?

• The codebase had numerous linting errors (ruff) including unused imports, variable shadowing, and invalid escape sequences
• There were 26+ mypy type errors throughout the codebase
• The code contained Python 2 compatibility code that is no longer necessary (Python 2 EOL was in 2020)
• Some files used outdated string encoding/decoding patterns
• CSV handling used inconsistent binary/text modes
• Type annotations were missing or incorrect in many places

• What is the new behavior (if this is a feature change)?

• All ruff linting errors are now fixed (unused imports, variable shadowing, invalid escape sequences)
• All mypy type errors are resolved (26+ errors reduced to 0)
• Python 2 compatibility code has been modernized to Python 3.10+ standards
• String handling now uses Python 3 native Unicode functionality
• CSV handling uses consistent and correct file modes
• Comprehensive type annotations added throughout the codebase
• Code is now fully compliant with modern Python standards

• Does this PR introduce a breaking change?
  No breaking changes to the public API. All core functionality has been tested and works correctly. The changes are internal modernization and code quality improvements that maintain full backward compatibility for users of the library.

Technical Details

Fixed Issues:

• Ruff errors: Unused imports, variable shadowing (e.g., csv variable shadowing csv module), invalid escape sequences in regex patterns
• MyPy type errors: Missing type annotations, incorrect return types, Optional handling, variable redefinition issues
• Python 2 compatibility: Removed unnecessary .decode() and .encode() calls, modernized string handling
• File handling: Fixed CSV binary vs text mode issues, resolved Traversable vs Path type conflicts
• Complex type inference: Fixed challenging type annotation issues in vector.py with nested tuple structures

Files Modified:

• epitran/_epitran.py: Fixed type annotations for word_to_tuples method
• epitran/vector.py: Resolved complex type inference issues with feature vectors
• epitran/bin/*.py: Fixed Python 2 compatibility code, CSV handling, type annotations
• epitran/*.py: Comprehensive type annotation updates, modernized string handling

Verification:

• All core classes tested and working correctly
• Epitran transliteration verified for multiple languages
• VectorsWithIPASpace functionality confirmed
• Type annotations match actual runtime behavior

The codebase now passes all linting checks and is ready for Python 3.10+ environments.