I would say interphase. Generally a cell should be in the interphase longer than in the mitotic phase where the chromatin is massively condensed and by this not suited for any ChIP-like experiments. A population of cells is typically never uniform in terms of cell cycle state. They are usually a heterogeneous consortium, even if being the same cell type or a cell line. The odds to catch all of them right in a mitotic stage is probably quite low. An exception would probably be if you synchronize your cells by e.g. serum starvation (all cells would have more/less the same cell cycle status at a given time point) and then you catch a time point where they are all in mitotic phase. I never heard or read that people commonly do this. An exception could be that a certain protein that you ChIP is known or under suspision to have very cell-cycle-dependent activity in terms of DNA binding, but this is thinking aloud, I do not have examples for this.