Use VectorOfArray for arrays in MIRK

[ErikQQY]

Part of #204

Use VectorOfArray to replace Vector{Vector}.

The reason why using VectorOfArray instead of DiffEqArray is that some arrays, like y₀ need extension when we halve the mesh and start a new nonlinear solving, and such extension for DiffEqArray is troublesome.

[github-actions]

Benchmark Results

	master	47aa7c0...	master/47aa7c0a6d1fb8...
Simple Pendulum/IIP/BoundaryValueDiffEq.MIRK2()	6.47 ± 0.19 ms	6.51 ± 0.19 ms	0.993
Simple Pendulum/IIP/BoundaryValueDiffEq.MIRK3()	2.33 ± 0.068 ms	2.33 ± 0.076 ms	0.998
Simple Pendulum/IIP/BoundaryValueDiffEq.MIRK4()	0.81 ± 0.029 ms	0.806 ± 0.029 ms	1.01
Simple Pendulum/IIP/BoundaryValueDiffEq.MIRK5()	0.857 ± 0.036 ms	0.85 ± 0.028 ms	1.01
Simple Pendulum/IIP/BoundaryValueDiffEq.MIRK6()	0.974 ± 0.029 ms	0.964 ± 0.03 ms	1.01
Simple Pendulum/IIP/MultipleShooting(10, Tsit5; grid_coarsening = false)	1.32 ± 0.099 ms	1.31 ± 0.097 ms	1.01
Simple Pendulum/IIP/MultipleShooting(10, Tsit5; grid_coarsening = true)	2.59 ± 0.19 ms	2.56 ± 0.17 ms	1.01
Simple Pendulum/IIP/MultipleShooting(100, Tsit5; grid_coarsening = false)	0.0347 ± 0.0047 s	0.0349 ± 0.0045 s	0.995
Simple Pendulum/IIP/MultipleShooting(100, Tsit5; grid_coarsening = true)	0.114 ± 0.0043 s	0.114 ± 0.0038 s	1
Simple Pendulum/IIP/Shooting(Tsit5())	0.189 ± 0.0057 ms	0.187 ± 0.0046 ms	1.01
Simple Pendulum/OOP/BoundaryValueDiffEq.MIRK2()	0.0396 ± 0.0014 s	0.0392 ± 0.0008 s	1.01
Simple Pendulum/OOP/BoundaryValueDiffEq.MIRK3()	11.5 ± 0.16 ms	11.9 ± 0.5 ms	0.97
Simple Pendulum/OOP/BoundaryValueDiffEq.MIRK4()	3.3 ± 0.06 ms	3.3 ± 0.061 ms	1
Simple Pendulum/OOP/BoundaryValueDiffEq.MIRK5()	3.38 ± 0.067 ms	3.4 ± 0.067 ms	0.994
Simple Pendulum/OOP/BoundaryValueDiffEq.MIRK6()	3.53 ± 0.079 ms	3.5 ± 0.077 ms	1.01
Simple Pendulum/OOP/MultipleShooting(10, Tsit5; grid_coarsening = false)	3.29 ± 0.74 ms	3.28 ± 0.85 ms	1
Simple Pendulum/OOP/MultipleShooting(10, Tsit5; grid_coarsening = true)	5.97 ± 1.2 ms	5.88 ± 1.2 ms	1.02
Simple Pendulum/OOP/MultipleShooting(100, Tsit5; grid_coarsening = false)	0.124 ± 0.01 s	0.122 ± 0.01 s	1.01
Simple Pendulum/OOP/MultipleShooting(100, Tsit5; grid_coarsening = true)	0.383 ± 0.0059 s	0.379 ± 0.0049 s	1.01
Simple Pendulum/OOP/Shooting(Tsit5())	0.776 ± 0.034 ms	0.774 ± 0.033 ms	1
time_to_load	5.69 ± 0.02 s	5.68 ± 0.032 s	1

Benchmark Plots

A plot of the benchmark results have been uploaded as an artifact to the workflow run for this PR.
Go to "Actions"->"Benchmark a pull request"->[the most recent run]->"Artifacts" (at the bottom).

[ChrisRackauckas]

lots of test failures here.