The reason I see for keeping both is the following.
Most commmon use case, the one having the security context performs
the FEC. In this case there are no issues with performing FEC first
and then encrypting before transmission.
The case I think will be less commonly used. But which still is
pretty useful case is when the media stream is first protected by
the media transmitter. Then one realize that FEC is needed, and
added by a node not trusteed with the security context. In that
case the FEC will be performed on the encrypted packets.
It would have been good to have only one case. However I have heard
of actually cases when this second use case is in use.