Lower tolerance for interpolate test

[leo-collins]

#4478

[connorjward]

Can you try pushing a couple of empty commits to this branch (git commit --allow-empty -m "Empty commit") so we can check that it doesn't happen over a couple of CI runs. Obviously give CI time to complete in between.

[connorjward]

Was this closed due to #4484? This would still be very helpful to merge into release as we are still seeing test failures there.

[leo-collins]

It didn't seem to fix the test failure. I'll try it again with an even lower tolerance

[connorjward]

If it doesn't work then I don't mind having stochastic failures in this one test until we merge main. Not worth spending too much time on.

I guess we could alternatively skip this test in release knowing that it is fixed in main.

[connorjward]

Let's close this. Merging main into release is going to happen soon.