Deterministic thermal drift tests

[rykerfish]

This PR is to address a comment left by a reviewer about testing the thermal drift. Mainly, it adds:

• a test that verifies convergence of RFDs to the divergence term computed deterministically.
• functionality to specify the RFD delta in the interface of NBody and DPStokes, as well as changes to specifying delta in units of hydrodynamicRadius

Here's a plot using DPStokes(single_wall) showing convergence of an RFD for a fixed particle configuration to the deterministic solution. It matches the expected $O(1/\sqrt{N})$ convergence:

I changed it to specify delta in terms of particle radius since it should have units of length and since we found delta needs to be small specifically compared to particle radius. Here's another plot showing an RFD for the same particle configuration using $a=0.1$ with DPStokes. If we just use $\delta=1e^{-3}$, the error plateaus early. Using $\delta=1e^{-3}a$ gives better convergence.

I also deleted the tests test_divM_is_zero and test_divM_matches_rfd as I think this test should supersede those two tests. It implicitly checks (to numerical error) that periodic/open geometries return 0 for divM since the deterministic divergence in those cases is numerically 0 (very small).

Final notes: NBody single_wall can converge somewhat slowly but runs quickly so I cranked the number of iterations it was allowed to go up to. DPStokes can reach an error plateau sometimes slightly before 1e-3, presumably since DPStokes is only designed for 2-3 digits of accuracy. For this reason, I used a lower tolerance to test convergence for DPStokes.

Let me know what you think!

[github-actions]

✅ Linter reported no issues

All Python files are correctly formatted with Black.

[github-actions]

✅ Linter reported no issues

All C/C++ files are correctly formatted with clang-format.

[RaulPPelaez]

LGTM! Thanks Ryker