refactor: evaluate compile-time template engine for UI

[devitway]

Problem

ui/templates.rs (1859 lines) and ui/components.rs (947 lines) generate HTML via format!() string concatenation with manual html_escape() calls. No compile-time verification of HTML structure or auto-escaping.

Risk

Missing a single html_escape() on user-controlled data (artifact names, versions) creates an XSS vector. Manual escaping is error-prone as the UI grows.

Options

Engine	Approach	Binary size	Compile-time safe
askama	Jinja2-like templates, compile-time	~0 runtime	Yes
maud	Rust macros as HTML	~0 runtime	Yes
tera	Runtime templates	+200KB	No
Current	format!() strings	0	No

Notes

This is a significant refactoring (2800+ lines). The current approach works and has no known XSS (all escape points are covered by tests). This is about long-term maintainability, not an urgent fix.

Evaluate whether the migration cost justifies the safety benefit given that UI is a secondary feature of NORA.