perf: improve perf of fill and Screen.init

Use @memset for initializing and filling a Screen.

src/Screen.zig

@@ -34,7 +34,7 @@ cursor_shape: Cell.CursorShape = .default,
 pub fn init(alloc: std.mem.Allocator, winsize: Winsize, unicode: *const Unicode) !Screen {
     const w = winsize.cols;
     const h = winsize.rows;
-    var self = Screen{
+    const self = Screen{
         .buf = try alloc.alloc(Cell, w * h),
         .width = w,
         .height = h,
@@ -42,9 +42,8 @@ pub fn init(alloc: std.mem.Allocator, winsize: Winsize, unicode: *const Unicode)
         .height_pix = winsize.y_pixel,
         .unicode = unicode,
     };
-    for (self.buf, 0..) |_, i| {
-        self.buf[i] = .{};
-    }
+    const base_cell: Cell = .{};
+    @memset(self.buf, base_cell);
     return self;
 }
 pub fn deinit(self: *Screen, alloc: std.mem.Allocator) void {

src/Window.zig

@@ -206,11 +206,18 @@ pub fn gwidth(self: Window, str: []const u8) usize {
 
 /// fills the window with the provided cell
 pub fn fill(self: Window, cell: Cell) void {
+    if (self.x_off == 0 and self.width == self.screen.width) {
+        // we have a full width window, therefore contiguous memory.
+        const start = self.y_off * self.width;
+        const end = start + (self.height * self.width);
+        @memset(self.screen.buf[start..end], cell);
+    } else {
+        // Non-contiguous. Iterate over rows an memset
         var row: usize = self.y_off;
         while (row < (self.height + self.y_off)) : (row += 1) {
-        var col: usize = self.x_off;
-        while (col < (self.width + self.x_off)) : (col += 1) {
-            self.screen.writeCell(col, row, cell);
+            const start = self.x_off + (row * self.screen.width);
+            const end = start + self.width;
+            @memset(self.screen.buf[start..end], cell);
         }
     }
 }