Replace sed post-processing in 2pydantic with gen-pydantic template override

[candleindark]

Summary

• Replace the sed -E 's,[a-z]+COLON,,g' post-processing in the 2pydantic hatch script with an enum.py.jinja override passed to gen-pydantic via --template-dir.
• The substitution now happens at the point where enum labels are emitted (via pv.label.split("COLON") | last), rather than as blind text replacement over the whole generated file.
• This removes the dependency on a system sed from the Pydantic generation pipeline, which makes the "Validate the data instances against the Pydantic models generated from the LinkML schema" task in Add LinkML behavior tests for required: False -> True slot_usage refinement #408 easier to implement (e.g. generating the Pydantic models in-process or in environments where shelling out to sed is awkward).

Why

The old pipeline was:

gen-pydantic --black dandischema/models.yaml | sed -E -e 's,[a-z]+COLON,,g' > dandischema/models_linkml.py

gen-pydantic emits enum members whose Python identifiers carry LinkML's namespace-prefix munging — e.g. dandiCOLONOpenAccess = "dandi:OpenAccess". The sed stripped the <prefix>COLON portion file-wide.

Doing this at the template level is more precise:

• it only touches the place where the prefix is known to be a munging artifact (the enum member label), not arbitrary file text;
• the intent is explicit in the template rather than hidden in a regex pipeline;
• it drops the runtime dependency on a system sed, which is useful for Add LinkML behavior tests for required: False -> True slot_usage refinement #408 (validating data instances against the generated Pydantic models) where we'd rather not have the generation step shell out.

Verification

Generated models_linkml.py two ways from dandischema/models.yaml (restored from the linkml-auto-converted branch):

1. old: gen-pydantic --black … | sed -E 's,[a-z]+COLON,,g'
2. new: gen-pydantic --black --template-dir tools/linkml_conversion_tools/pydantic_templates …

diff of the two outputs is empty — byte-identical. All 116 COLON occurrences in the raw gen-pydantic output are preceded by a lowercase prefix, so split("COLON") | last is equivalent to the sed on this schema.

Test plan

• Compared old vs new output on dandischema/models.yaml from linkml-auto-converted — byte-identical.
• On the linkml-auto-converted branch (or one stacked on top), run hatch run linkml-auto-converted:2pydantic and confirm dandischema/models_linkml.py is unchanged from before this PR.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 97.83%. Comparing base (08e2d11) to head (23666ae).

Additional details and impacted files

@@                  Coverage Diff                  @@
##           linkml-conversion     #410      +/-   ##
=====================================================
+ Coverage              96.67%   97.83%   +1.16%     
=====================================================
  Files                     20       19       -1     
  Lines                   2436     2407      -29     
=====================================================
  Hits                    2355     2355              
+ Misses                    81       52      -29     

Flag	Coverage Δ
unittests	97.83% <ø> (+1.16%)	⬆️

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

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.