Allow reporting timing data for several pipelines at once (from any Detectors or DistributionMaker instance) ...

[thehrh]

consistently propagate value of profile to any contained Pipeline instances and enable customisable number format (making use of the very powerful utils.format.format_num function).

Note that Pipeline.get_outputs(), Pipeline.setup(), and Pipeline.run() can now be separately profiled, so one does not have to manually add up the timings from the individual services within a pipeline and any overhead from computations in addition to Pipeline.run() becomes visible.

Example output from a call to new DistributionMaker.report_profile() with default args:

Pipeline: neutrinos
- setup:        Total time (s): 61.584, n calls: 1
- run:          Total time (s): 3.825, n calls: 2, time/call (s): mean 1.913, max. 3.535, min. 0.291
- get_outputs:  Total time (s): 4.080, n calls: 2, time/call (s): mean 2.040, max. 3.547, min. 0.533
Individual services:
data simple_data_loader
- setup:    Total time (s): 0.010, n calls: 1
- compute:  Total time (s): +0.000, n calls: 1
- apply:    Total time (s): 0.017, n calls: 2, time/call (s): mean 0.008, max. 0.011, min. 0.006

....

Pipeline: muons
- setup:        Total time (s): 0.006, n calls: 1
- run:          Total time (s): 2.621, n calls: 2, time/call (s): mean 1.310, max. 2.619, min. 0.001
- get_outputs:  Total time (s): 2.624, n calls: 2, time/call (s): mean 1.312, max. 2.622, min. 0.002
Individual services:
data simple_data_loader
- setup:    Total time (s): +0.000, n calls: 1
- compute:  Total time (s): +0.000, n calls: 1
- apply:    Total time (s): +0.000, n calls: 2, time/call (s): mean +0.000, max. +0.000, min. +0.000

...

Resolves #815