syslog flooded by 1.36.35

[iconnor]

I believe I have recreated the problem. One of my cameras does it. The first packet that comes out has pts unset, which triggers the line setting last_dts=-1. This results in strangeness. Clearly the -1 is wrong. Just not sure yet what the right thing to do is.

Be patient, we'll get it fixed for 1.36.36.

[iconnor]

I've looked hard at the code and as near as I can tell, all that duration calculation code is just garbage.

So I've replaced it all with one line. Will test for a bit, but I'm for more confident in the new code than in the old and it seems to be working so far.