Expose subtree processing from the morph_stats api

[eleftherioszisis]

Allow to process heterogeneous subtrees for morph_stats analysis.

Examples usage:

neurom.apps.morph_stats.extract_dataframe(moprh, config, use_subtrees=True)
neurom.apps.morph_stats.extract_stats(moprh, config, use_subtrees=True)

#975

[codecov-commenter]

Codecov Report

❗ No coverage uploaded for pull request base (v4@006cafc). Click here to learn what that means.
The diff coverage is n/a.

@@          Coverage Diff           @@
##             v4     #1034   +/-   ##
======================================
  Coverage      ?   100.00%           
======================================
  Files         ?        35           
  Lines         ?      2464           
  Branches      ?         0           
======================================
  Hits          ?      2464           
  Misses        ?         0           
  Partials      ?         0           

[adrien-berchet]

LGTM but why is the default set to False? I think the results from the old behavior are a bit hard to understand for a user who probably expects the results given by use_subtrees=True by default, WDT?

[eleftherioszisis]

LGTM but why is the default set to False? I think the results from the old behavior are a bit hard to understand for a user who probably expects the results given by use_subtrees=True by default, WDT?

Default is False because this is the normal behavior as one would expect using entire trees of single types, i.e. the legacy behavior. I don't think we should break backward compatibility given that it is not necessary here.

Edit: However, making heterogeneous subtree processing the new standard is an enticing idea that would make things simpler.

[eleftherioszisis]

@adrien-berchet may I merge this?

[adrien-berchet]

Yep, let's keep this default value