Some issues with BVDAE solvers

[ErikQQY]

With the merge of #217, the BVDAE solvers are available now🎉🎉🎉, but the internal need to be further refactored, here are some possible aspects for improvement:

• The boundary conditions definition is different with the MIRK and shooting, actually there is also a difference between MIRK and Shooting, which reminds me of getting Boundary conditions should always use solution object #208 fixed.
• The fixed points specification in BVDAE solvers, while MIRK, FIRK, and shooting methods just state condition points in boundary conditions definition, Ascher needs to explicitly state where the points are for the evaluation.
• The internal side conditions evaluation is bad, we need to refactor that.
• The recursive flatten and unflatten of arrays in all BVP solvers are being used everywhere, maybe we can directly bake them into RecursiveArrayTools.jl.
• Some variables are not properly named, need to refactor these.
• The jacobian of boundary conditions is inconvenient to use, we need to change that.
• More......