Flesh out string byte handling a bit:

- Get rid of public byteAt(_) method on strings. It's redundant and
  longer than .bytes[_].
- Implement bytes.count natively so it's O(1).

test/core/string_byte_sequence/count.wren

@@ -0,0 +1,15 @@
+// Simple.
+IO.print("".bytes.count) // expect: 0
+IO.print("123".bytes.count) // expect: 3
+
+// UTF-8.
+// Bytes:
+//        123456789
+// Chars: sø mé ஃ
+IO.print("søméஃ".bytes.count) // expect: 9
+
+// Null bytes.
+IO.print("\0\0\0".bytes.count) // expect: 3
+
+// Invalid UTF-8.
+IO.print("\xef\x00".bytes.count) // expect: 2