fix MacOS CI again

[akhra]

tested re-running a passing test on #279, it fails now; looks like something upstream changed again.

[akhra]

(tentative assignment, feel free to reject)

[akhra]

@mpilgrem I've bumped version on master to 0.1.25.0. the remaining in-flight PR will require 0.2.0.0, might as well batch in the non-breaking changes first. should we formally cut that version now, or would you like to examine the MacOS CI failure more first?

[mpilgrem]

@akhra, from what I have read, this could be a real thing on a GitHub macOS runner - not a passing glitch. I am going to try some experiments ...