Sprucing rate tests failed in nightlies

Report of a crash/segfault with Moore

Brief summary of encountered problem

The benchmark test Moore_spruce_rate_and_test crashed since June 1st, some of the previous tests are fine (e.g. test for lhcb-master-mr.4580).

Dataset

All files of upgrade-minbias-hlt2-persistreco-output in TestFileDB, which are output MDF files from a series of Hlt2 persistreco lines.

Moore version used (or nightly, as appropriate)

Any nightly build since June 1st. An example is lhcb-master.1701. The platform is x86_64_v3-centos7-gcc11-opt+g. Its log file of the Moore run is under

/eos/lhcb/storage/lhcbpr/www/UpgradeRateTest/RateTest_lhcb-master.1701_Moore_spruce_rate_and_size_x86_64_v3-centos7-gcc11-opt+g_2022-06-02_10:56:53_+0200/RateTest.x86_64_v3-centos7-gcc11-opt+g.16t.1j.1000000e.0.log

Option file used to execute Moore

The test can be launced locally by (under stack)

./Moore/run DBASE/PRConfig/scripts/benchmark-scripts/Moore_spruce_rate_and_size.sh

Which was tested before it was merged in late May.

The corresponding option file is $HLT2CONFROOT/options/sprucing/spruce_all_lines_analytics.py.

Additional information

cc @abertoli @nskidmor