[RTADPA BW Tests] Take account of compression changes at Hlt2/Sprucing

We use os.stat(fpath).st_size to extract the size of the streams in Bytes. (For Hlt2 we also run zstd -5 {fpath} to compress the Hlt2 output and extract that filesize too).

Then in line-and-stream-rates we calculate the compression factor and use this when calculating the avg evt size / dstdata size per line and per stream.

• Hlt2: zstd compressed fsize / default fsize
• Spruce: default fsize / sum(size(rawbanks)) for file.

cc @rjhunter @shunan

Locally evaluated changes

Evaluated ['these changes', 'these changes + forcing compression_factor=1'] with a reduced set of Hlt2 lines (to reduce load on local testing).

Can see:

hlt2_bandwidth

Compression factor set to 1:

Compression factor evaluated correctly:

spruce_bandwidth

We recover reasonable looking output from the spruce test instead of the ~3x larger output seen on the UpgradeRateTest pages on master.

ToDo:

• Minimal changes to account for Hlt2 compression changes
• How do we want to handle Spruce compression changes?
  • https://mattermost.web.cern.ch/lhcb/pl/ju3eq9z6u38otjty8wmsh854xa
  • Thanks @nskidmor @cburr
• Discuss how best to report this on the pages.
  • !406 (comment 7839253) + !406 (comment 7841309)