feat: benchmarking setup (#85)

* switch to merged input files
* pass custom schema to ServiceX (requires coffea>=0.7.18)
* split coffea pre-processing into separate step
* track and save metrics when running coffea
* add I/O testing option
* add option to scale fraction of files read
* add updated data sizes for merged files

Author: alexander-held

.gitignore

@@ -25,6 +25,7 @@ workshops/agctools2022/statistical-inference/input
 
 # CMS ttbar
 analyses/cms-open-data-ttbar/workspace.json
+analyses/cms-open-data-ttbar/metrics
 
 # dask
 dask-worker-space/