Use scifi_v6 sequence for SMOG2 CI test

The SMOG2 throughput test running in the CI is currently using the hlt1_pp_default sequence (CI config script). However, since the sample contains scifi v6 raw banks, the hlt1_pp_scifi_v6 sequence should be used instead. This will likely decrease the throughput and should then be more similar to the test we run on the other scifi v6 minbias sample. Note that the dumped geometry for the SMOG sample should also be passed. This is likely wrong at the moment as well. This can be done similarly as here, but using smog as geometry (which points to this directory).

I propose to add a physics efficiency test for the SMOG2 sample in addition. Is there a pppHe sample for a physics channel?

When processing hlt1_pp_scifi_v6_validation on the SMOG2 sample and using the geometry dumped here, I get a segmentation fault from prefix_sum_ut_hits. However, when dumping the SMOG2 sample into MDF files as input for Allen, I re-dumped the geometry. Using the newly dumped geometry, there is no problem, so likely there was an update in the UT geometry causing the segfault.

The dumped SMOG2 MDF files and geometry can be found on eos: /eos/lhcb/wg/rta/WP6/Allen/mdf_input/SMOG2_pppHe/ and on the GPU development server: /scratch/allen_data/mdf_input/SMOG2_pppHe/

Action items for this:

• Use hlt1_pp_scifi_v6 sequence for SMOG2 throughput test
• Update dumped geometry in Allen repository for SMOG2 sample
• Pass path to correct dumped geometry for SMOG2 throughput test
• Add a physics efficiency test for the SMOG2 sample