feat(core): Jinja2 .template processing and .append file convention

[ulises-jeremias]

Summary

Add CNA-parity file processing when loading template/extension layers:

1. .template files — process with Jinja2 ({{ variable }}), strip the suffix from the output path.
2. .append files — append content to an existing matching file instead of overwriting.

Variables come from scaffold context (projectName, customOptions answers from cpa.config.json, etc.).

CNA mapping

CNA: EJS .template + .append conventions in cna-templates / loaders (docs/AUTHORING.md in cna-templates).

CPA today: copy-only merge in create_python_app_core.loaders (copy_tree). Documented as a known limitation in cpa-templates docs (ARCHITECTURE.md, AUTHORING.md).

Acceptance criteria

• .template files are rendered with Jinja2 and written without the .template suffix
• Undefined variable policy documented (strict vs empty) and covered by tests
• .append appends to the destination file when it exists; creates it when missing
• Layer order still applies: template → addons → extend
• Scaffold context exposes at least projectName and cpa.config.json custom option answers
• Unit tests for .template, .append, and interaction with overwrite
• Update cpa-templates AUTHORING docs once behavior ships (or open a follow-up there)

Out of scope (follow-ups)

• [bracket]/ directory renaming
• .if-* conditional file inclusion

Blocked by

none

Blocks

Better extension authoring in cpa-templates (today extensions must ship complete overlay files).

Reference

• Parity matrix: knowledge/create-python-app-parity-matrix.md (workspace)
• CNA: https://github.com/Create-Node-App/cna-templates/blob/main/docs/AUTHORING.md
• CPA loaders: packages/create-python-app-core/src/create_python_app_core/loaders.py