parse_to_param(): Rework previous fix (OSS-Fuzz)

Invoking strlen() on a non NULL-terminated buffer is equally bad,
possibly even worse.  Instead, simply rely on the @EnD parameter being
correct, which is an assumption already made by the current code of the
function.

Fixes OSS-Fuzz#41073

parser/parse_to.c

@@ -479,8 +479,7 @@ static inline char* parse_to_param(char *buffer, char *end,
 
 parse_error:
 	LM_ERR("unexpected char [%c] in status %d: <<%.*s>> .\n",
-	    tmp < end? *tmp : *(buffer+strlen(buffer)-1),status,
-	    (int)(tmp-buffer), ZSW(buffer));
+	    tmp < end? *tmp : *(end-1),status, (int)(tmp-buffer), ZSW(buffer));
 error:
 	if (param) pkg_free(param);
 	free_to_params(to_b);
@@ -817,8 +816,7 @@ static inline char* _parse_to(char* buffer, char *end, struct to_body *to_b,
 
 parse_error:
 	LM_ERR("unexpected char [%c] in status %d: <<%.*s>> .\n",
-	    tmp < end? *tmp : *(buffer+strlen(buffer)-1),
-	    status, (int)(tmp-buffer), buffer);
+	    tmp < end? *tmp : *(end-1), status, (int)(tmp-buffer), buffer);
 error:
 	first_b->error=PARSE_ERROR;
 	free_to_params(first_b);