Fix overflows with malformed lesskey lines (#234)

If lines are malformed it is possible to trigger out of boundary
read accesses during parsing.

Carefully handle the pointer increments to not move behind the
terminating nul byte.

How to reproduce:

python -c 'print(1022*" "+"\\")' > lesskey-1.txt
python -c 'print(1022*" "+"^")' > lesskey-2.txt
python -c 'print(1021*" "+"\\k")' > lesskey-3.txt

Open these files with lesskey, compiled with -fsanitize=address.

lesskey_parse.c

@@ -180,6 +180,18 @@ char_string(buf, ch, lit)
  return buf;
 }
 
+/*
+ * Increment char pointer by one up to terminating nul byte.
+ */
+ static char *
+increment_pointer(p)
+ char *p;
+{
+ if (*p == '\0')
+ return p;
+ return p+1;
+}
+
 /*
  * Parse one character of a string.
  */
@@ -255,7 +267,7 @@ tstr(pp, xlate)
  case '1': ch = SK_F1; break;
  default:
  parse_error("invalid escape sequence \"\\k%s\"", char_string(buf, *p, 0));
- *pp = p+1;
+ *pp = increment_pointer(p);
  return ("");
  }
  *pp = p+1;
@@ -274,7 +286,7 @@ tstr(pp, xlate)
  * Backslash followed by any other char 
  * just means that char.
  */
- *pp = p+1;
+ *pp = increment_pointer(p);
  char_string(buf, *p, 1);
  if (xlate && buf[0] == CONTROL('K'))
  return tstr_control_k;
@@ -284,13 +296,13 @@ tstr(pp, xlate)
  /*
  * Caret means CONTROL.
  */
- *pp = p+2;
+ *pp = increment_pointer(p+1);
  char_string(buf, CONTROL(p[1]), 1);
  if (xlate && buf[0] == CONTROL('K'))
  return tstr_control_k;
  return (buf);
  }
- *pp = p+1;
+ *pp = increment_pointer(p);
  char_string(buf, *p, 1);
  if (xlate && buf[0] == CONTROL('K'))
  return tstr_control_k;