Switch default propagator to Dormand Prince 78

After recent tests, it's very obvious that the Dormand Prince integration scheme is more stable than an RK89, and should be faster (less stages and one order less).

The specific example that showed the problem is using the Ruggiero control with an adaptive step size in an RK89: this shows that the control vector is regularly NaN, showing instability of the computation of the state. In fact, with an RK89 and Ruggiero control, the step size must be very appropriately selected to ensure that the control is not NaN. With a Dormand Prince 78, this problem completely vanishes!

Moreover, for propagation, the energy bleed is about 3 times as high with an RK89 than a Dormand Prince 78, both in 1 day propagation and 30 day propagation!

One day

Note that the test compares the final state with an RK89 from GMAT, hence the larger errors!

running 1 test
[RK89] ==> energy_conservation absolute errors with RK89 val state      energy bleed = 3.824851546596619e-11
4e-11   2e-9    3e-9    3e-12   2e-12   1e-12
[DP78] ==> energy_conservation absolute errors with RK89 val state      energy bleed = -1.2441603303159354e-11
2e-8    5e-7    7e-7    7e-10   4e-10   3e-10
test mission_design::orbitaldyn::energy_conservation ... ok

test result: ok. 1 passed; 0 failed; 0 ignored; 0 measured; 130 filtered out; finished in 1.40s

Thirty days

running 1 test
[RK89] ==> energy_conservation absolute errors with RK89 val state      energy bleed = 1.1477432337869686e-9
[DP78] ==> energy_conservation absolute errors with RK89 val state      energy bleed = -3.8112801803436014e-10
test mission_design::orbitaldyn::energy_conservation ... ok

test result: ok. 1 passed; 0 failed; 0 ignored; 0 measured; 130 filtered out; finished in 23.88s

For validation purposes, all of the validation shall still use the rk89 call.