Bug fix and improvement in TrackMate export

[qin-yu]

Hey @JoOkuma,

@tischi and I encountered a small but critical and elusive bug, fixed by this PR.

Fig. 1. TrackMate. Before: no tracks | After: tracks | After: details

Fig. 2. Mastodon. After: tracks and details

The current test for exporting tracks to TrackMate solely verifies the inclusion of all spots in the output, neglecting tracking semantics. In this PR, I choose not to implement an additional test that merely converts the XML back, since the essential validation comes from the actual opening and viewing of successful tracks. I revised the function's name, though I suspect it has not been used for a while.

Before Fix

TrackMate launches with error:

Some errors occurred while reading file:

Unknown spot ID: 2000008
Could not find the detector element in file.
Could not find the tracker element in file.
Could not find the feature analyzer element.
Could not find the display-settings element. Returning user defaults.

Mastodon throws error only:

Exception in thread "Thread-8" java.lang.NullPointerException
	at org.mastodon.mamut.io.importer.trackmate.TrackMateImporter$Import.<init>(TrackMateImporter.java:429)
	at org.mastodon.mamut.io.importer.trackmate.TrackMateImporter.readModel(TrackMateImporter.java:236)
	at org.mastodon.mamut.launcher.MastodonLauncher.lambda$importMaMuT$18(MastodonLauncher.java:504)
	at java.lang.Thread.run(Thread.java:748)

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 71.42857% with 2 lines in your changes are missing coverage. Please review.

❗ No coverage uploaded for pull request base (main@09807e8). Click here to learn what that means.

Files	Patch %	Lines
ultrack/core/export/trackmate.py	71.42%	2 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main      #90   +/-   ##
=======================================
  Coverage        ?   85.41%           
=======================================
  Files           ?       71           
  Lines           ?     3916           
  Branches        ?        0           
=======================================
  Hits            ?     3345           
  Misses          ?      571           
  Partials        ?        0           

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

[JoOkuma]

Hi @qin-yu and @tischi, thanks once again for your contribution.

I think I figured out why we did not observe this bug before; we always assumed the id matches the data frame index. And this is not always true, especially depending on how you load/save your data frame.

Could you check if my suggestions fix your problem while avoiding the line

parent_spot_id = tracks_df.index[tracks_df['id'] == parent_id][0] # uniqueness checked above

It fixed the example that I generated on my side.

[JoOkuma]

LGTM! Thanks @qin-yu