well i seem to have solved the problem at least partially myself
in the TCP_in file in the LWIP source, i disabled the piece of code in the ACK handling routine where it accepts the TCP window update value and puts the value into code, this stopped the random crashing on two concurrent window updates
however, i think that by doing this i disabled a part of TCP flow controll, so performance might not benefit, the normal numbers ive seen are similar to how fast it was with the code in original form, but it still isnt recommendable i think to try this
if i have anymore news ill post here to log the problem/solution for anyone who happens upon the same troubles