Effective spin sometimes missed by generate_all_posterior_samples

In the latest version of PESummary, it seems that all spin components must be known in order to get generate_all_posterior_samples to derive \chi_{\mathrm{eff}} (see [source])

This means, for instance, GWTC-1 data release files cannot derive \chi_{\mathrm{eff}}, despite having spins available in polar coordinates.

It would be very helpful to add support for deriving \chi_{\mathrm{eff}} from other parameterizations. My suggestion is to hard-code the two following cases: ($\chi_1, \cos\theta_1, \chi_2, \cos\theta_2) and (\chi_{1z}, \chi_{2z}$). In fact, relaxing the if-statement I linked above to only require ["spin_1z", "spin_2z"] may fix this in many cases.