Makefile for HILE using cray toolchain, building for CPU.

[ursg]

Also includes building of boost program_options as part of build_libraries.sh

Not sure if that is the way we want to encourage people to build it, but I guess it is at least useful to have the commands for it in the libraries script.

[ursg]

In testpackage evaluation, the vlsvdiffs of some tests (but not all?) segfault:

----------
running Flowthrough_x_inflow_y_outflow 
----------
--------------------------------
 ref-time | new-time | speedup |
--------------------------------
 3.385      3.956      0.855662
-------------------------------------------
 variable                                     | absolute diff | relative diff |
-------------------------------------------
Comparing file /wrk-vakka/group/spacephysics/proj/CI/hile-uganse/_work/vlasiator/vlasiator/testpackage/run_2024.11.04_09.38.38/Flowthrough_x_inflow_y_outflow/bulk.0000001.vlsv against reference
 proton/vg_rho_0                                3.49e-10        3.49e-16
 proton/vg_v_0                                  3.49e-10        6.48e-16
 proton/vg_v_1                                  1.16e-10        3.84e-16
 proton/vg_v_2                                  1.04e-11        0.94
 fg_b_0                                         0               0
 fg_b_1                                         0               0
 fg_b_2                                         0               0
 fg_e_0                                         5.42e-20        3.37e-16
 fg_e_1                                         0               0
 fg_e_2                                         8.13e-20        3.49e-16
VLSV file timestamps match.
----------
Comparing file /wrk-vakka/group/spacephysics/proj/CI/hile-uganse/_work/vlasiator/vlasiator/testpackage/run_2024.11.04_09.38.38/Flowthrough_x_inflow_y_outflow/bulk.0000002.vlsv against reference
srun: error: x3000c0s14b1n0: task 0: Segmentation fault (core dumped)
 proton/vg_rho_0                                                
srun: error: x3000c0s14b1n0: task 0: Segmentation fault (core dumped)
 proton/vg_v_0                                                  
srun: error: x3000c0s14b1n0: task 0: Segmentation fault (core dumped)
 proton/vg_v_1                                                  
srun: error: x3000c0s14b1n0: task 0: Segmentation fault (core dumped)
 proton/vg_v_2                                                  
srun: error: x3000c0s14b1n0: task 0: Segmentation fault (core dumped)
 fg_b_0                                                         
srun: error: x3000c0s14b1n0: task 0: Segmentation fault (core dumped)
 fg_b_1                                                         
srun: error: x3000c0s14b1n0: task 0: Segmentation fault (core dumped)
 fg_b_2                                                         
srun: error: x3000c0s14b1n0: task 0: Segmentation fault (core dumped)
 fg_e_0                                                         
srun: error: x3000c0s14b1n0: task 0: Segmentation fault (core dumped)
 fg_e_1                                                         
srun: error: x3000c0s14b1n0: task 0: Segmentation fault (core dumped)
 fg_e_2                                                         
awk: cmd. line:1: BEGIN{print (!= 0.0)?1:0}
awk: cmd. line:1:              ^ syntax error
awk: cmd. line:1: BEGIN{print (!= 0.0)?1:0}
awk: cmd. line:1:                     ^ syntax error
VLSV file timestamps match.
----------

Is this just a result of carrington-reference-data being compared with HILE results? Shouldn't our testpackage be resilient to that?

[ursg]

Also, the testpackage is much slower (by a factor ~4) than on carrington, and eventually runs into a timeout.
This smells like our old friend, the core placement bug.