Tupek/tr core mfem subspace

[tupek2]

The indent here is mostly to get away from requiring PETSc to use the 'subspace' option for the trust-region solver, and to have additional header files for the trust-region implementation so the equations_solver.cpp does not keep getting larger. There were some additional performance optimizations, such as reducing the total dot products and the number of parallel reductions, saving off subspace results when doing linesearching (as the subspace does not change), tuned some defaults based on a suite of benchmarks that I will add in a follow on PR, and some other small adjustments.

• Remove PETSc/SLEPc dependency from trust-region subspace solver.
• Replace PETSc reduced solve with MFEM/LAPACK implementation.
• Cache reduced subspace projections across trust-radius retries.
• Extract Steihaug-Toint CG into private implementation helper.
• Keep public entry through NonlinearSolver::TrustRegion.
• Use one batched global dot-product path for trust-region math.
• Add previous accepted steps to subspace basis, default 2.
• Add tuned trust-region defaults from benchmark runs.
• Add MFEM subspace and Steihaug-Toint CG tests.
• Guard LAPACK subspace tests on MFEM_USE_LAPACK.
• Add shallow-arch snap-through physics coverage.