[whitescreendev]

This documentation update clarifies how string-based similarity functions handle Unicode text.

It documents that:

• string-based functions operate on Rust char values rather than UTF-8 bytes;
• canonically equivalent strings can have different scalar-value sequences;
• Unicode normalization may be needed before comparison;
• user-perceived characters can contain multiple Unicode scalar values;
• grapheme-cluster segmentation may be appropriate when one visible character should count as one edit.

The implementation and API are unchanged. This is a documentation-only update.

An additional technical reference is included for Unicode normalization, grapheme segmentation, and text transformations in edit-distance processing.