Remove expensive equilibrium deepcopying

[timothy-nunn]

NOTE: this PR requires FusionComputingLab/freegs4e#47 and should probably be pinned to version 0.12+ of freegs4e.

Removes expensive deepcopies of equilibrium objects inside the evolutive loop.

The following table shows the % of the runtime for a 60 iteration example evolutive run that is spent deep copying objects.

	Linear	Non-linear
main (10615f0)	$11.17$%	$5.82$%
This PR	$3.25$%	$3.82$%

Overall, this results in a reduction in runtime spent in stepping.nlstepper (linear evolutive example): going from $21.12$% on main to $16.96$% on this branch.

I have compared history arrays of the coil currents, o-points, elongation, triangularity, squareness, separatrix area, and separatrix length between main and this PR. There is a 0 MAE between the non-linear example histories on main and in this PR, and a 0 MAE between the linear histories on main and in this PR.

I have confirmed that example 01, 02, 03, 04, 05, 09, and 10 can run on this branch and produce the same results as on main (the example are all non-deterministic because the RNG is not seeded but they are qualitatively similar to examples as run on main)

[kpentland]

Changes look great Tim, thanks!

I've run the examples and also tested examples 6a/b to ensure everything still works for MAST-U equilibria. Seems to be working nicely.

We'll update the freegs4e version, get the tests run, and then merge in.