What makes a histogram different from an ordinary bar chart is that its bars sit directly next to each other with no gaps in between. That visual choice is not arbitrary, it reflects the fact that the data being plotted flows continuously along the scale, with each class interval picking up exactly where the previous one left off.
Discrete data does not have that property, since discrete values are separate and countable, so discrete frequency distributions are shown with a bar diagram that deliberately leaves gaps between bars to show the categories are distinct. Cumulative frequency, meanwhile, is not plotted as bars at all, it is plotted as a smooth running-total curve called an ogive. Relative frequency is really just describing what is on the vertical axis, proportions instead of raw counts, and does not by itself say anything about whether the underlying data is continuous.
Since only continuous data has adjoining, gapless class intervals, a histogram, with its touching bars, can only be properly constructed for a continuous frequency distribution.
That makes continuous frequency distribution the correct answer.