Fix misleading PG_RE_THROW commentary

The old verbiage indicated that PG_RE_THROW is optional, which is not
really true.  This has confused many people, so it seems worth fixing.

Discussion: https://postgr.es/m/20190206160958.GA22304@alvherre.pgsql

src/include/utils/elog.h

@@ -273,7 +273,9 @@ extern PGDLLIMPORT ErrorContextCallback *error_context_stack;
  *
  * (The braces are not actually necessary, but are recommended so that
  * pgindent will indent the construct nicely.)  The error recovery code
- * can optionally do PG_RE_THROW() to propagate the same error outwards.
+ * can either do PG_RE_THROW to propagate the error outwards, or do a
+ * (sub)transaction abort. Failure to do so may leave the system in an
+ * inconsistent state for further processing.
  *
  * Note: while the system will correctly propagate any new ereport(ERROR)
  * occurring in the recovery section, there is a small limit on the number