FastInterpolations breaks unit tests and doc building

[jhalpern30]

#137 was merged in with a few existing bugs that break both the unit tests and the documentation building. I think I have a quick fix to the documentation in #150, but the unit tests are still breaking, so that PR also comments out the broken unit tests for now, but these will need to be fixed. I am going to document what's broken in this issue.

MORE TO COME

[jhalpern30]

#150 has now fixed the documentation build issues and the Solovev/CHEASE unit tests errors. The only remaining error/commented out unit test in the multi-n Solovev full run-through, where I am getting:

[ Info: Running Solovev ideal multi-n example
DCON START
----------------------------------
[ Info: Forcing pest coordinate jacobian exponents: power_*
Equilibrium file: /Users/jakehalpern/Github/JPEC/test/test_data/regression_solovev_ideal_example_multi_n/sol.toml
Generating Solovev equilibrium inputs with:
   mr=128, mz=128, ma=128
   e=1.6, a=0.33, r0=1.0
   q0=1.9, p0fac=1.0, b0fac=1.0, f0fac=1.0
   Magnetic axis found at R = 1.00000, Z = 0.00000
   Inboard separatrix found at R = 0.5830951894666458.
   Outboard separatrix found at R = 1.2884098727169977.
┌ Warning: Automatic dt set the starting dt as NaN, causing instability. Exiting.
└ @ OrdinaryDiffEqCore ~/.julia/packages/OrdinaryDiffEqCore/GMkz9/src/solve.jl:645
┌ Warning: NaN dt detected. Likely a NaN value in the state, parameters, or derivative value caused this outcome.
└ @ SciMLBase ~/.julia/packages/SciMLBase/RPW6n/src/integrator_interface.jl:618
Full DCON runs: Error During Test at /Users/jakehalpern/Github/JPEC/test/runtests_fullruns.jl:12
  Test threw exception
  Expression: begin
    #= /Users/jakehalpern/Github/JPEC/test/runtests_fullruns.jl:13 =#
    DCON.Main(ex2)
    #= /Users/jakehalpern/Github/JPEC/test/runtests_fullruns.jl:14 =#
    true
end
  One of the first solution vector norms unorm(1,1) = 0

This is relatively concerning, as at first glance it seems like the splines are breaking some physical quantity in the code such that it can't be integrated anymore. This could be as simple as something multi-n related was not integrated properly, but will need to be looked into.

[logan-nc]

The multi-n bug was introduced from overzealous optimization attempts that initialized the fundamental matrices using undef instead of zeros. Since the multi-n only calculates block diagonal terms we had a bunch of nans in the integrations.

This is solved in 9c6011e

[logan-nc]

The docs is a separate and mysterious issue. I am moving it to a devoted issue #154