Interesting question!
Uranium isotopes can decay by alpha decay, beta decay, or spontaneous fission. 235U and 236U both decay almost entirely by alpha decay. What matters is not just the stability of the parent nucleus but the stability of the parent *relative* to the daughter, which is measured by the alpha-decay energy E. It's generally observed that the log of the alpha-decay half-life is approximately a linear function of E^-1/2. For these two parents we have E=4.7 MeV and 4.6 MeV respectively. That would suggest that they would have nearly the same alpha-decay half-lives
However, it is always found that odd nuclei have slower alpha-decay rates than their even-even neighbors. The ratio of these quantities is called the hindrance factor. The hindrance factor is apparently mainly due to angular momentum. In 236U->232Th, we have a 0+ even-even nucleus going to another even-even nucleus with spin-parity 0+. But in 235U->231Th the ground states are 7/2- and 5/2+. The change in parity actually violates a parity selection rule, so I guess the 235U must decay to some excited state of 231Th, not the ground state. But anyway there is probably a difference in angular momentum, which has to be accounted for by the alpha particle. As an alpha particle with angular momentum L tunnels out through the Coulomb barrier, it also experiences a centrifugal barrier proportional to L(L+1). I think this centrifugal barrier is what causes the much longer half-life compared to 236U.
Note that the short half-lives of 237 and 239U are because they are unstable with respect to beta decay.