This seems like an excellent fit for BigQuery... The dataset can be shared publicly in BigQuery as well.
BigQuery is a big overkill for a dataset that fits in memory. Colab + Pandas is the way to go unless you want to sit around twiddling thumbs while BigQuery shuffles the data around.
It's definitely overkill, but it does run fast on small datasets. (almost any query would return in a few seconds at this size)
Also, I find it's SQL dialect and UI very nice for simple stuff too.