I think this time I have the issue nailed down. fingers crossed.
I tried running multiwii - differnt 328p - differnt code - same hardware - same issue
This is extremely weird as I use ppm in on my multicopters and it has run fine for years, so deff a hardware issue.
I grabbed the wing from the plane this will be going in (4 9g digital servos) and tested both yours and the multiwii code, both work flawlessly.
To confirm what on my test bed was causing issues I swapped one servo on the wing for first the esc on the test bed. Work like gold, then swapped the esc for the servo (as the test bed is a rock crawler it is a full size all metal gear beast of a servo, but it is analog....) once that was in the system all hell broke lose once again.
This is confirmed both by visually seeing the servos twitch and from the multiwii gui/prints in your code
So as far as time goes it was that servo causing all the issues, somehow creating a bucketload of noise in the system. A ferrite ring may weed out the issue, or giving it its own power.
This brings me to another scenario, my original code that used servo timer 2 may have never been broken in the first place. When I removed the esc from that setup it worked fine - possibility that specific pieces of hardware (or combination of) cause can cause issues.
If I get the time I may try to sort that issue, it would be nice to have a solution now that I know the problem.
Thankyou for your help and I hope this discovery of mine can shed light on any future problems you may run into.
Will post back with ether A)sucses story of project or B)another issue