I think the major error in the purported proof is that the line graph has all its cycles having length not congruent to 1 m odmod $d+1$ is non trivial(there are cycles having length greater than $d+1$), or is altogether false in some cases ( as shown by the example given by @bof below in the comment).