grep: option to use GNU regex matching instead of POSIX one.

This fixes problems with NULs in files being scanned, but
 costs +800 bytes. The same can be done to sed (TODO).

libbb/xregcomp.c

@@ -27,6 +27,6 @@ void FAST_FUNC xregcomp(regex_t *preg, const char *regex, int cflags)
 {
 	char *errmsg = regcomp_or_errmsg(preg, regex, cflags);
 	if (errmsg) {
-		bb_error_msg_and_die("xregcomp: %s", errmsg);
+		bb_error_msg_and_die("bad regex '%s': %s", regex, errmsg);
 	}
 }