What is found in your attachment is
[tex]\int_0^\pi (cos(2x)+ cos(2t))y(t)dt= ky(x)[/tex]
not quite what you originally wrote.
We can write that as
[tex]cos(2x)\int_0^\pi y(t)dt+ \int_0^\pi cos(2t)y(t)dt= ky(x)[/tex]
As elibj123 pointed out, both of those integrals are NUMBERS that, of course, depend on y(x).
Let [tex]Y_1= \int_0^\pi y(t)dt[/tex]
and [tex]Y_2= \int_0^\pi cos(2t)y(t)[/tex].
Now the equation reads simply [itex]ky(t)= Y_1 cos(2x)+ Y_2[/itex] and we can solve for y(t) just by finding the two numbers [itex]Y_1[/itex] and [itex]Y_2[/itex].
Multiply both sides of [itex]ky(x)= Y_1 cos(2x)+ Y_2[/itex] by cos(2x) and integrate from 0 to [itex]\pi[/itex]:
[tex]k\int_0^\pi cos(2x)y(x)dx= Y_1 \int_0^\pi cos^2(2x) dx+ Y_2\int_0^\pi cos(2x)dx[/tex]
Of course, [itex]\int_0^\pi cos(2x)y(x)dx[/itex] is the same as [itex]\int_0^\pi cos(2t)y(t)dt[/itex] which we have called [itex]Y_2[/itex]. The other two integrals do not involve y(x) and so can be integrated. Rather than do them for you I am going to write [itex]\int_0^\pi cos^2(2x)dx= A[/itex] and [itex]\int_0^\pi cos(2x)= B[/itex] (although that second one ought to be obvious!).
Now, our equation is [itex]kY_2= AY_1+ BY_2[/itex] or [itex]AY_1+ (B- k)Y_2= 0[/itex].
If we simply integrate [itex]ky(x)= Y_1 cos(2x)+ Y_2[/itex] itself from 0 to [itex]\pi[/itex] we get
[tex]k\int_0^\pi y(x)dx= Y_1 \int_0^\pi cos(2x)dx+ Y_2\int_0^\pi dx= BY_1+ \pi Y_2[/tex]
That is the same as [itex]kY_1= BY_1+ \pi Y_2[/itex] or [itex](B-k)Y_1+ \pi Y_2[/itex].
That is, we can solve for y(x) by solving the pair of numerical equations
[itex](B- k)Y_1+ \pi Y_2= 0[/itex] and [itex]AY_1+ (B- k)Y_2= 0[/itex]
where A and B are given by the integrals above.
Of course, like any eigenvalue equation, those are satified by the "trivial" solution [itex]Y_1= Y_2= 0[/itex]. Eigenvalues are values of k for which there exist non-trivial solutions. Non-trivial solutions for homogeneous systems of equations occur when the determinant of the coefficient matrix, here [itex](B- k)^2- A\pi[/itex], is 0.