v1.3.0

What's new

• Add real-valued, double-precision temperature types — celsius_real, kelvin_real, and fahrenheit_real — whose count() reads directly in scale degrees. Intended use is casting an exact reading to a real-valued type for display or floating-point computation, e.g. temperature_cast<celsius_real>(millicelsius(22500)).
• Their std::formatter specializations render the degree value with the scale suffix (22.5°C) and honor the standard floating-point format spec ({:.1f} → 22.5°C).
• Include <version> before the std::format feature-test check, so the formatter block is enabled in standalone (non-IDF) builds where nothing pulls in __cpp_lib_format first.

Full Changelog: v1.2.2...v1.3.0