Add unit tests for TextUnit S-expression generation

[jwalsh]

The project needs comprehensive unit tests for the TextUnit.to_s_expr method in the proust_reader.py file. These tests should verify proper S-expression generation with various nesting levels, metadata handling, and special characters. The tests should cover:

• Different node types (paragraphs, sentences, phrases, words)
• Proper handling of Unicode characters in the text
• Metadata serialization and deserialization
• Nesting hierarchies with multiple levels
• Edge cases like empty units or very large text blocks

This will help ensure the stability of the S-expression output format as the project evolves.