First cleanup of TestFileDB, dropped 144 entries for 5K lines

For a week, all slots running with TestFileDB have benefited from !469 and thus I have a good idea of which entries are used and not.

For a quick summary, there are 352 entries to be dropped out of 560 !

This MR is a first drop of 144 entries believed (by me) to be rather older than the average (measure being the data in the entry name or the place in the file). I'm not so sue how to test this. I think I'll apply the following steps :

• test in lhcb-head
• if ok, also test in 2012, 2016 and 2018 slots for a variety of cases
• if ok, go for all slots

@clemenci @bcouturi @msaur @gcorti @nskidmor @cburr for info, validation and comments