refactor: shared Unicode display-width utility with grapheme cluster support

Consolidate two inconsistent codepoint_width implementations from
Screen.cpp and EventEditor.cpp into nocal/tui/unicode.hpp. The unified
table covers all CJK, Hangul, emoji, symbol, and combining mark ranges.
display_width and fit_text handle grapheme clusters (ZWJ sequences,
skin-tone/hair modifiers, regional indicator pairs, variation selectors)
so calendar text with emoji or East Asian characters renders correctly.

include/nocal/tui/unicode.hpp

@@ -0,0 +1,35 @@
+#pragma once
+
+#include <cstddef>
+#include <cstdint>
+#include <string>
+#include <string_view>
+
+namespace nocal::tui {
+
+struct DecodedCodepoint {
+    std::uint32_t value;
+    std::size_t length;
+};
+
+// Decode one UTF-8 codepoint at position `at` in `text`.
+// Returns {0xfffd, 1} on error. Caller must ensure at < text.size().
+[[nodiscard]] DecodedCodepoint decode_codepoint(std::string_view text, std::size_t at) noexcept;
+
+// Terminal display width of a single codepoint: 0 (zero-width/combining), 1, or 2.
+// Uses Unicode East Asian Width property with complete range tables.
+[[nodiscard]] int codepoint_width(std::uint32_t value) noexcept;
+
+// Display width of a string. Handles grapheme clusters:
+// - Emoji ZWJ sequences → width 2
+// - Emoji + skin tone / hair modifiers → width 2
+// - Regional indicator pairs (flags) → width 2
+// - Combining marks → absorbed by base
+[[nodiscard]] int display_width(std::string_view text);
+
+// Truncate to `width` display cells. If `ellipsis` is true and the text is
+// truncated, appends U+2026. Right-pads with spaces to reach `width`.
+// Never breaks a grapheme cluster.
+[[nodiscard]] std::string fit_text(std::string_view text, int width, bool ellipsis = true);
+
+} // namespace nocal::tui