Can you provide a minimal working example, where spinor expressions are not simplified?

As for the Gordon identity, this is not a transformation that always leads to simpler expressions, so it is not useful to apply it everywhere. It is of course handy in the evaluation of vertex functions, where you want to extract the different form factors, but in other calculations it might not make much sense to trade $\gamma^\mu$ for $\sigma^{\mu \nu}$. Furthermore, it is very easy to apply Gordon identity via a replacement rule, see how it is done in

https://github.com/FeynCalc/feynhelpers/blob/master/Examples/QCD/NRQCDVertexMatchingInTheTwoQuarkSectorOneLoop.m

and

https://github.com/FeynCalc/feynhelpers/blob/master/Examples/QED/QEDRenormalizationOneLoop.m

Can you provide a minimal working example, where spinor expressions are not simplified?

As for the Gordon identity, this is not a transformation that always leads to simpler expressions, so it is not useful to apply it everywhere. It is of course handy in the evaluation of vertex functions, where you want to extract the different form factors, but in other calculations it might not make much sense to trade $\gamma^\mu$ for $\sigma^{\mu \nu}$. Furthermore, it is very easy to apply Gordon identity via a replacement rule, see how it is done in

https://github.com/FeynCalc/feynhelpers/blob/master/Examples/QCD/NRQCDVertexMatchingInTheTwoQuarkSectorOneLoop.m

and

https://github.com/FeynCalc/feynhelpers/blob/master/Examples/QED/QEDRenormalizationOneLoop.m

EDIT 2020: Gordon identities have been recently added to the development version via a new function GordonSimplify

Evaluating

SpinorUBar[a, b].GA[c].SpinorU[a, b] // GordonSimplify // FCE

returns

(Spinor[Momentum[a], b, 1].Spinor[Momentum[a], b, 1] FV[a, c])/b