Move solve for EnsembleProblem and _concrete_solve_adjoint to SciMLBase

[jClugstor]

Checklist

• Appropriate tests were added
• Any code changes were done in a way that does not break public API
• All documentation related to code changes were updated
• The new code follows the
  contributor guidelines, in particular the SciML Style Guide and
  COLPRAC.
• Any new documentation only uses public API

Additional context

Currently solve for Ensemble and WeightedEnsemble problems is in DiffEqBase. However there can be ensembles of NoninearProblems as well as DiffEqProblems. So this is part of the process of removing the DiffEqBase dependency from NonlinearSolve.

[jClugstor]

@ChrisRackauckas I think this needs to go in, then this SciML/DiffEqBase.jl#1201 to remove it from DiffEqBase. Then I think both SciML/DiffEqBase.jl#1197 and SciML/NonlinearSolve.jl#669 will be very close to finished.

[jClugstor]

@ChrisRackauckas for some reason when solve(prob::EnsembleProblem... is defined in both SciMLBase and DiffEqBase it causes some Dual type check tests to error, which is what the DiffEqBase core test failures are from. Maybe something to do with the method being overridden?

Anyway, the PR to remove these from DiffEqBase is here SciML/DiffEqBase.jl#1201, but the tests for both of these won't pass without the other.

[jClugstor]

@ChrisRackauckas should I split this PR up or is this ok?

[ChrisRackauckas]

Downstream isn't setup with !isdefined yet?