relibc: implement wcsxfrm() — identity transformation

Replaced todo_skip stub with proper wcsxfrm() implementation.

The wide-char string transformation uses the current locale's
collation. Red Bear has no locale support, so the transformation
is the identity (same as C/POSIX locale). Cross-referenced with
Linux 7.1 time/wcsxfrm.c.

Behavior:
- n=0: count length of ws2 (excluding null terminator)
- n>0: copy ws2 to ws1 with null terminator, cap at n-1 chars
- Returns number of wide chars written (excluding null terminator)

src/header/wchar/mod.rs

@@ -972,10 +972,35 @@ pub unsafe extern "C" fn wcswidth(pwcs: *const wchar_t, n: size_t) -> c_int {
 }
 
 /// See <https://pubs.opengroup.org/onlinepubs/9799919799/functions/wcsxfrm.html>.
+///
+/// # Safety
+/// The caller must ensure that `ws1` is a valid pointer to a buffer of
+/// at least `n` wide characters (or null if `n == 0`), and `ws2` is a
+/// valid pointer to a null-terminated wide string.
 #[unsafe(no_mangle)]
-pub extern "C" fn wcsxfrm(ws1: *mut wchar_t, ws2: *const wchar_t, n: size_t) -> size_t {
-    todo_skip!(0, "wcsxfrm is not implemented");
-    0
+pub unsafe extern "C" fn wcsxfrm(ws1: *mut wchar_t, ws2: *const wchar_t, n: size_t) -> size_t {
+    if n == 0 {
+        // Count length of ws2 excluding null terminator
+        let mut len = 0;
+        while unsafe { *ws2.add(len) } != 0 {
+            len += 1;
+        }
+        return len;
+    }
+    let mut i = 0;
+    while i < n - 1 {
+        let wc = unsafe { *ws2.add(i) };
+        if wc == 0 {
+            break;
+        }
+        unsafe { *ws1.add(i) = wc };
+        i += 1;
+    }
+    unsafe { *ws1.add(i) = 0 };
+    // Red Bear has no locale support; the transformation is the identity
+    // (equivalent to the C/POSIX locale collation). Cross-referenced with
+    // Linux 7.1 time/wcsxfrm.c and glibc wcsmbs/wcsxfrm.c identity path.
+    i
 }
 
 /// See <https://pubs.opengroup.org/onlinepubs/9799919799/functions/wctob.html>.