Threads and progress

[bendavid]

Updated singularity image eliminates most spurious threads from Tensorflow Lite by adding a dedicated patch (previously one spurious thread per RDF-thread per tflite helper, now reduced to one thread in total for tflite)

Updated env variables in narf eliminate an addition ncores threads from Eigen.

Updated singularity image contains a general update of package versions, but also ROOT 6.30.02 (plus our patches for improved PyROOT debugging which aren't upstream yet)

narf has been adapted to implement a variation on the Progress Bar feature (but in a way which is compatible with RunGraphs)

Zero physics changes expected (but maybe some numerical differences/reshuffling depending)

[cippy]

The crash occurring in the histmakers might have been caused by a transient excessive usage of memory, David (and myself at some point) was running something almost saturating the cern machine memory

[bendavid]

No this is a technical problem with tensorflow. Already testing a fix.

[bendavid]

From in-progress CI logs:

begin lumi loop
[Total elapsed time: 0:22m  processed files: 100 / 100  processed evts: 992 / 992]   
end lumi loop
dataPostVFP 0.16036740900000002
begin event loop
|====>               |   |==================> |   [Elapsed time: 5:00m  processing file: 314 / 1163  processed evts: 68022000 / 68947393  2.27e+05 evt/s]   
|======>             |   |==================> |   [Elapsed time: 10:00m  processing file: 424 / 1163  processed evts: 112391000 / 113048301  1.87e+05 evt/s]   
|=======>            |   |==================> |   [Elapsed time: 15:00m  processing file: 497 / 1163  processed evts: 134375000 / 135285574  1.49e+05 evt/s]   
|=========>          |   |==================> |   [Elapsed time: 20:00m  processing file: 603 / 1163  processed evts: 158462000 / 159195216  1.32e+05 evt/s]   

[bendavid]

ci output is 1:1 identical