Implement batched serial laswp

[yasahi-hpc]

This PR implements laswp function, which is needed for getrf PR.

Following files are added:

1. KokkosBatched_Laswp_Serial_Impl.hpp: Internal interfaces with implementation details
2. KokkosBatched_Laswp.hpp: APIs
3. Test_Batched_SerialLaswp.hpp: Unit tests for that

Detailed description

This performs a series of row interchanges on a general rectangular matrix.

• A: (batch_count, m, n) or (batch_count, m) On entry, the M-by-N matrix or the length M vector. The row interchanges will be applied to the matrix of column dimension N. On exit, the permuted matrix or vector.
• IPIV: (batch_count, m) The pivot indices; for 0 <= i < m, row i of the matrix was interchanged with row IPIV(i).

Parallelization would be made in the following manner. This is efficient only when A is given in LayoutLeft for GPUs and LayoutRight for CPUs (parallelized over batch direction).

Kokkos::parallel_for('laswp', 
    Kokkos::RangePolicy<execution_space> policy(0, n),
    [=](const int k) {
        auto sub_ipiv = Kokkos::subview(m_ipiv, k, Kokkos::ALL());
        auto sub_a = Kokkos::subview(m_a, k, Kokkos::ALL(), Kokkos::ALL());
        KokkosBatched::SerialLaswp<typename ParamTagType::direct>::invoke(sub_ipiv, sub_a);
    });

Tests

1. Make a random matrix from random A and ipiv. Apply Laswp to A while preparing a reference Ref from A which is permuted by ipiv. Then confirm A == Ref.