Yeah, the naming in arviz.summary is not ideal. Not 100% sure why (might have used ETIs at some point or aim to support both or just be a confusion). The hdi function doesn’t use this naming and sets as coordinates 'lower', 'higher'. Not sure if we already have an issue but a PR would be welcome. Thinking out loud, to keep the column names short for the summary dataframe, we might want to use hdi_low and hdi_up?
Note now there is a stats_focus argument (last example in the docstring) and naming for ETI should continue to have the specific probabilities.