[FEATURE]: Clearer, cleaner string template / pattern generation

[josh-seidel-db]

Summary

Redesign and clarify the string template / pattern generation mechanism (e.g. TemplateGenerator and template-based text generation) so the template mini-language is clearer to author and cleaner to maintain.

Background

dbldatagen supports template-driven text generation, where templates use escape sequences and special meta-characters to express patterns (digits, letters, words, ranges, repetition, alternation, etc.). As the template language has grown, the syntax and the code that parses it have become harder to read, document, and extend.

Problems

• Template syntax (escapes / meta-characters) is not always intuitive and is easy to get wrong.
• Edge cases — escaping literals, repetition counts, optional and alternation groups — are under-documented.
• The parsing/handling code is harder to maintain and extend than it should be.

Proposed direction

• Define a clear, documented grammar for the template / pattern language.
• Improve error messages for malformed templates.
• Separate parsing from generation to improve testability.
• Provide a compatibility / migration path so existing templates continue to work.
• Expand the documentation with a complete reference table and worked examples.

[mvanhorn]

Opened #454 to address the documentation deliverable here: a complete template/pattern grammar reference table and worked examples in docs/source/textdata.rst, with the TemplateGenerator docstring kept in sync as the source of truth. It covers the full set of meta-characters and character classes, escaping and the escapeSpecialChars semantics, alternation, and worked examples, and corrects the d/D rows (d is any digit 0-9, D is non-zero 1-9). The grammar/parser redesign portion of this issue is intentionally left out of that PR.