Change of Tracer changes the results

[nabobalis]

Describe the bug

For tests:

• test_field_line_polarity
• test_footpoints

If you replace the Python Tracer with the one from StreamTracer, the polarity and the number of solar_footpoint change breaking both tests.

On the surface I would expect both traces to give the same output.

Question is, do older version of streamtracer agree or not
Also is this a problem? Should they give the same output?

[STBadman]

i'll look into this - it seems possible there could be some differences from numerical issues if the discretization is different between the tracers, but a big change like a reversed polarity is surprising to me at first.

[nabobalis]

It it possible its just an artifact of the way the test is setup.

I did try streamtracer from 2.0 to 2.2 and they all had the same result.
I just tried the FORTRAN version (1.2.0), that's the same as well.