In classical Spanish poetry, Góngora’s Soledades provide an excellent example both of a masterful use of complex tonal configurations and of how a conventional, strictly rhythmic metrics is bound to miss this quality, however easily perceptible to a reader: the often admirable Navarro is forced to conclude, given his assumptions, that Góngora’s metrical skill does not go beyond “a level many others did not exceed either”. A view of metrics that does not lead to this kind of a conclusion would seem to me to be preferable, everything else being equal.
--Carlos Piera, "REPHRASING LINE-END RESTRICTIONS," emphasis added.
Understatement differs from litotes* in that it is not (necesarily) the negation of a negative, but simply a tonal shading. Look at this example from the brilliant poet and linguist Carlos Piera. He cites a quite outrageous opinion by the great Spanish prosodist Tomás Navarro Tomás. Instead of taking outraged issue with it, he states his discrepancy in measured, scholarly tones. This is effective because it works at a literal level (yes, it is preferable, mutatis mutandis) or as a deliberate rhetorical flourish. You feel the writer holding back, and you can supply the outrage (f any) yourself.
This kind of understatement, then, is useful for very strong disagreement.
*Litotes could be defined as a specialized case of understatement, but not all understatements are cases of litotes.