That's not really a feature we have. We do support SSRC rewriting, but
it's done as a 1-to-1 mapping, i.e. if there are two SSRCs incoming then
you will get two (different) SSRCs outgoing.
There is a technical complication to what you're asking for, which is
that merging two SSRCs into one also necessarily requires merging the
sequence numbers and the timestamps, as well as information contained in
RTCP reports. So it's not entirely trivial.
You can probably achieve a similar result by enabling the "audio player"
feature, which consumes all received audio and produces a single
consistent output RTP stream. However this requires decoding and
re-encoding all audio, plus the added scheduling overhead, so the extra
CPU cost might be an issue.
Cheers