fix: stop nickname truncation cutting a character in half

truncateNickname compared and cut with String.length and take, both of which
count UTF-16 code units. A name whose 15th unit is the high surrogate of an
emoji lost its low surrogate, leaving a lone U+D83D that renders as a tofu box
wherever the name is shown - the peer list, the geohash people list, and every
inline mention.

Step back to the last grapheme boundary at or before the limit instead, so a
truncated name is always whole. The limit itself is unchanged, so an ASCII
name truncates exactly as before.

app/src/main/java/com/bitchat/android/ui/ChatUIConstants.kt

@@ -1,9 +1,19 @@
 package com.bitchat.android.ui
 
+import java.text.BreakIterator
+
 /**
  * UI constants/utilities for nickname rendering.
  */
 fun truncateNickname(name: String, maxLen: Int = com.bitchat.android.util.AppConstants.UI.MAX_NICKNAME_LENGTH): String {
-    return if (name.length <= maxLen) name else name.take(maxLen)
-}
+    if (name.length <= maxLen) return name
 
+    // The limit counts UTF-16 code units, so cutting at it can land inside a
+    // surrogate pair and leave a lone surrogate that renders as a tofu box, or
+    // split a ZWJ sequence into its parts. Step back to the last grapheme
+    // boundary at or before the limit so what is shown is always whole.
+    val boundaries = BreakIterator.getCharacterInstance()
+    boundaries.setText(name)
+    val end = boundaries.preceding(maxLen + 1)
+    return if (end == BreakIterator.DONE) "" else name.substring(0, end)
+}