You can get all the roots for any trinomial by using the reversion theorem depending on the radius of convergence. Surpsingly, so such solution exits online for the general trinomial.
so for example $x^5-2x^2+z=0$ we have two roots with a Euclidian norm of less than one and three roots greater than one.
we would have two partitions:
$f(x)=x^5-2x^2$ which gives three roots
you you would apply it around the region of $2^{1/3}$ which gives a real root and two complex ones. For small z, one can check that these are close to the actual answer
or
$f(x)=-2x^2+z$ and gives the remaining two
This bypasses the resolvent method, which results in multiple hypergeometric series for each root. You only need one hypergeometric function for each root.
this combines this
https://arxiv.org/pdf/0910.2957.pdf
with
https://en.wikipedia.org/wiki/Bring_radical
the first paper gives the formula for the small roots (14,15)
the wiki page (Glasser's derivation) will give the others
It's very hard to find good summaries about how to apply the Lagrange inversion to actually derive these. If anyone can prove these formulas it would be great. You have to make some transformations so $x^5-2x^2+z=0$ becomes $1+y^2+y^5(1/z)(-z/2)^{5/2}$ and the root of the original equation is $y(-z/2)^{1/2}$. It can be done.