If time to the singularity is considered finite it can never require an infinite degree of fine tuning, but I can see how this could conflict with a spatially infinite geometry together with a global CP as I mentioned in my previous post.Hardly. The cosmological principle holding to infinity requires an infinite degree of fine-tuning: how did the universe, out to infinite distances, know to be the same density in all locations, with the appropriate time-slicing?
It's rather like the horizon problem, but expanded to infinite distances instead of merely being required to hold in our visible universe.
Another way of stating the problem is to look at the classic model of inflation. If inflation were extended infinitely into the past, then inflation could easily explain a global cosmological principle. However, we know that can't be the case: extending inflation infinitely into the past also requires infinite fine-tuning: inflation predicts a singularity somewhere in the finite past, and the further back you try to push that singularity, the more fine-tuning you need. And if inflation can only be extended a finite distance back into the past, then it isn't possible for the universe as a whole to have reached any sort of equilibrium density, as if you go far enough away, you'll eventually reach locations that have always, since the start of inflation, been too far for light to reach one another. Any regions of the universe that lie beyond this distance aren't likely to be remotely close to one another in density.
Of course, this argument is based upon the assumption that a simplistic model of inflation is true, but the argument is reasonably-generic among most inflation models.
But I just can't understand how exactly the horizon problem came to be considered a problem in the first place(unless the above mentioned conflict was also evident at the time), the three possible space geometries allowed by the FRW model are both geometrically and physically homogeneous by definition, so if one is going to use this model one is assuming that homogeneity and shouldn't worry about causal contact.