Profile code

[transientlunatic]

This PR adds tests to profile the code more effectively.

[transientlunatic]

-------------------------------------------------------  ------------  ------------  ------------  ------------  ------------  ------------  ------------  ------------  ------------  ------------
                                                   Name    Self CPU %      Self CPU   CPU total %     CPU total  CPU time avg     Self CUDA   Self CUDA %    CUDA total  CUDA time avg    # of Calls
-------------------------------------------------------  ------------  ------------  ------------  ------------  ------------  ------------  ------------  ------------  ------------  ------------
                                       model_likelihood        51.52%     693.273ms        99.88%        1.344s        1.344s       0.000us         0.00%     278.159ms     278.159ms             1
                                  aten::linalg_solve_ex         0.00%      54.000us        23.12%     311.050ms     155.525ms       0.000us         0.00%     231.637ms     115.819ms             2
                                 aten::_linalg_solve_ex         0.02%     223.000us        23.11%     310.964ms     155.482ms       0.000us         0.00%     231.637ms     115.819ms             2
                              aten::linalg_lu_factor_ex         0.96%      12.924ms         9.46%     127.295ms      42.432ms     172.744ms        62.10%     173.101ms      57.700ms             3
                                       aten::linalg_inv         0.01%      70.000us        19.12%     257.323ms     257.323ms       0.000us         0.00%     159.835ms     159.835ms             1
                                    aten::linalg_inv_ex         0.02%     205.000us        18.36%     247.055ms     247.055ms       0.000us         0.00%     159.832ms     159.832ms             1
                                  aten::linalg_lu_solve         0.03%     367.000us        13.69%     184.271ms      92.135ms       3.025ms         1.09%      95.305ms      47.653ms             2
                          aten::linalg_solve_triangular         0.09%       1.188ms         9.22%     124.011ms      31.003ms      92.038ms        33.09%      92.038ms      23.009ms             4
void cutlass::Kernel<cutlass_80_tensorop_d884gemm_64...         0.00%       0.000us         0.00%       0.000us       0.000us      84.977ms        30.55%      84.977ms       2.360ms            36
                                     aten::linalg_solve         0.00%      14.000us         5.69%      76.628ms      76.628ms       0.000us         0.00%      71.864ms      71.864ms             1
-------------------------------------------------------  ------------  ------------  ------------  ------------  ------------  ------------  ------------  ------------  ------------  ------------
Self CPU time total: 1.346s
Self CUDA time total: 278.159ms