Input: a line steps W1 W2 … (bin widths to try), then a CSV with a column duration_min.
Print n=16, min 4, max 31, mean 16.9, median 11.5. For each width w: bins start at ⌊min/w⌋·w and are half-open [a, a+w); use ⌊(max − start)/w⌋ + 1 bins (edges start + w * np.arange(nbins + 1)) and np.histogram; print step 5: [0,5):1 [5,10):5 … and peaks: 2 -> bimodal or multimodal (or unimodal), where a peak is a non-empty bin larger than its left neighbour (or first) and at least as large as its right neighbour (or last).
Finally build an Altair histogram with the last width (alt.Bin(step=w), y = count()) and print altair bin: {'step': 5.0} | y: count.