fix symmetry reduction for force

[stevengj]

Fixes #129, following discussion in #3161.

The fix is to add an optional second "conjugate" component cc to a volume_list whose phase (conjugated) is included in symmetry reductions via reduce. For the force calculations, cc is set according to the normal_direction.

cc @oskooi — this PR is just a rough draft right now, and could use a test (e.g. copied from #3161).

[stevengj]

I copied over the test from the other PR. Would be good to double-check that it fixes the original test case in #129.

[oskooi]

The fix proposed in this PR (ceff208) does not produce the expected result of zero force for the test involving an Ez dipole in 2d vacuum. (The test is based on the one provided in #129 and transcribed from Scheme to Python. The simulation layout is shown below.) Running the test script with the --with-symmetry flag (mirror symmetry in $x$ and $y$) produces an incorrect force of 0.07380557. The result is -1.0482879755219221e-17 without symmetries (which is correct).

Additionally, using only a mirror symmetry in $x$ produces the correct result . However, a $y$ mirror symmetry produces 0.07380557 which is incorrect. This suggests there is a bug in the $y$ mirror-symmetry case.