Hi, sorry I wasn't entirely clear! I meant that size=None only has to be set whenever the expected results would be affected by the change. In most cases the size=10_000 default acts the same as size=None where it matters.

Thanks to your changes I could also easily see that we were (potentially) doing a lot of extra work in unit tests, retrieving possibly very many results while only caring whether or not we received 50 (or some other number). I also updated those tests to retrieve a more reasonable number of results.

Thanks for the help! You should easily be able to commit my suggested changes (or, if you think this can cause issues, reject them). After that's done I think the PR is good to merge 👍 (see note on progress_rst in comment below)