add completed_at field (+run_stats) to TaskDoc as root fields

[tsmathis]

@esoteric-ephemera, this should be enough to get completed_at in as a root level field, right?

There were already kwargs of completed_at=calcs_reversed[0].completed_at, in the .from_structure calls, but the field isn't showing up currently

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 88.84%. Comparing base (de70eb1) to head (02838e4).

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1116   +/-   ##
=======================================
  Coverage   88.83%   88.84%           
=======================================
  Files         114      114           
  Lines       10627    10629    +2     
=======================================
+ Hits         9441     9443    +2     
  Misses       1186     1186           

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

[esoteric-ephemera]

@tsmathis that should be good - also worth noting that run_stats and include_structure are two other optional fields that are automatically populated by TaskDoc.{from_directory, from_vasprun}. We should probably make these default top-levels too right?

Also tweaked one of the builder tests to better account for when no POTCAR library is preset (like CI)

[tsmathis]

I can't really find anywhere that include_structure actually has an effect?

And the run_stats would be okay to include I guess since they were in TaskDocument. Will be just another field that we'll have to drop when parsing though.

There is also differing behavior for from_directory and from_vasprun for the run stats field, should they be the same?

[esoteric-ephemera]

I'm not sure what the purpose of include_structure is - it's not used anywhere in emmet or atomate2 and it's just a bool. I'd be tempted to change it to a property of TaskDoc that returns whether TaskDoc.structure is None

from_directory and from_vasprun have slightly different outputs - the from_vasprun method doesn't populate as many fields because it doesn't have access to things like volumetric objects. Their signatures are the same tho

[esoteric-ephemera]

Hey @utf, do you recall what the purpose of TaskDoc.include_structure is? We'd like to remove it from the TaskDoc.from_directory and TaskDoc.from_vasprun methods, since it's not defined in the base schema of TaskDoc.

[utf]

IIRC it was an option as to whether the structure was stored alongside the metadata. It was originally part of StuctureMetadata.from_structure. I don't remember it being a field though.

[tsmathis]

That was the impression we got based on the name, but we haven't found any place in either emmet or atomate2 where include_structure has an effect for TaskDoc.

Is this a behavior you'd like to see functioning for TaskDoc?

[utf]

As long as the structure is always stored in the task doc that's fine. Sounds like it's safe to remove.

[tsmathis]

Okay, sounds good @utf, thanks for the input.

@esoteric-ephemera, lets follow up separately on the include_structure behavior and make sure no regressions are introduced.

That aside, going to merge this in unless there is anything else you'd like to see in these current changes?