Make RB random again

[hay-k]

Summary

RB circuits were not random enough. This PR makes them random enough.

Two examples

before: http://login.qrccluster.com:9000/Szg1hROpSDCJFjBKXUtPtQ==/
after: http://login.qrccluster.com:9000/uf9sLKHkSBSjm9iZruRP0g==/

before: http://login.qrccluster.com:9000/oGx4JpRpTZGONzR1nsBzFA==/
after: http://login.qrccluster.com:9000/iGTKwIO8Rhewk0dY2lJsQQ==/

Comments

• Because of this bug the results were highly dependent on parameters used to run the experiment (depths, nither, seed). If you were lucky you could even get proper RB output, but most of the time the quality of data degraded quickly with increasing depth.
• If you ran RB only for short depths (e.g. like this one), which a lot of people do apparently, you would think the experiment works fine and the data is fine. But increasing the depth reveals the issue.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 97.48%. Comparing base (ed9fb0b) to head (4704913).
Report is 17 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #995      +/-   ##
==========================================
- Coverage   97.50%   97.48%   -0.03%     
==========================================
  Files         123      123              
  Lines        9714     9714              
==========================================
- Hits         9472     9470       -2     
- Misses        242      244       +2     

Flag	Coverage Δ
unittests	97.48% <100.00%> (-0.03%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
...qibocal/protocols/randomized_benchmarking/utils.py	95.80% <100.00%> (ø)

... and 1 file with indirect coverage changes

[Edoardo-Pedicillo]

Thanks @hay-k