Fix bug in when reconstructing a 3D field from thetaMode data

[AngelFP]

In #337, a bug was introduced that causes the 3D reconstruction to fail if Nr > max_res_transv/2. This is because Nr is no longer being updated after reducing the radial resolution of the Fcirc array. This PR adds back the lines of code that were updating this value.

[AngelFP]

Somehow the Windows tests have failed, but it does not seem to be related to the changes in this PR. It cannot find the openPMD test data (RuntimeError: Found no valid files in directory ./example-2d/hdf5/.)

[ax3l]

CI:

Collecting openpmd-api
  Downloading openPMD_api-0.14.5-cp39-cp39-win_amd64.whl (1.7 MB)
     ---------------------------------------- 1.7/1.7 MB 36.2 MB/s eta 0:00:00
...
Please check that this is the path to the openPMD files.(valid files must have one of the following extensions: ['json', 'bp'])

Means that my latest wheel build (openPMD/openPMD-api#1286) for Windows did not compile-in HDF5 properly. Will fix: openPMD/openPMD-api#1305

[AngelFP]

Thanks for having a look at it!

[RTSandberg]

Thanks for catching and fixing this!

[ax3l]

With openPMD/openPMD-api#1305 merged, Windows pip HDF5 should work again 🎉 I'll restart CI for it.

[AngelFP]

Thank you!