Only average over trains dimension in Scan.bin_by_steps()

[takluyver]

This came up at FXE: if you have an array with multiple values per train, such as a spectrum, you naturally want an average spectrum per scan step, but scan.bin_by_steps(arr) reduces it down to a single number per scan step instead.

This could theoretically break code that relies on it collapsing extra dimensions, but I'm not sure when you'd want that, and it's very easy to do further averaging on the results from this if you need to. So I hope we can just consider it a bug and fix it. 🤞

cc @bermudei

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 73.93%. Comparing base (65ca7c8) to head (57ab6f3).
Report is 8 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #269   +/-   ##
=======================================
  Coverage   73.93%   73.93%           
=======================================
  Files          23       23           
  Lines        2996     2996           
=======================================
  Hits         2215     2215           
  Misses        781      781           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[JamesWrigley]

Won't this also remove the extra dimension names and such? It'd be nice to preserve those if possible. This should also have a test.

[takluyver]

You're quite right, I hadn't read the code carefully enough. 👍

[JamesWrigley]

LGTM!

[takluyver]

Thanks!