fix: restore required semantics for a bare Link[T]

[simontaurus]

Reverts the Link[T] optionality change that shipped in 2.0.0. Everything else from #162 stays: the emitter conformance work is unaffected, and emitted schemas still validate against OO-LD 1.0.0-rc.5.

#162 carried feat!: a bare Link[T] annotation is optional and closed #159 automatically. #159 is not settled:

• 19 Sep, simontaurus: "Preliminary decission: Option 2"
• 21 Sep, MatPoppFHG: proposes the opposite shape, bare Link[T] stays required with Optional[Link["Person"]] = None for the optional case, and raises the open question of what happens when one link in a.father.father.father cannot be resolved

A preliminary decision plus an unanswered counter-proposal is not a decision. The library adopted one side of it.

Reopening #159. The chain-resolution question Matthias raised is the part that decides this: if a bare Link[T] is optional, guard-free chaining stops being guaranteed, which is the property the annotation exists to provide.

Surgical: only 1eb6a57 is reverted, leaving the four emitter fixes and the tests from the same branch. 751 passed, 10 skipped, 1 xfailed.

[github-actions]

Release preview

No version bump from the current commits (stays at v1.1.0). Use conventional commit types (feat, fix, ...) to trigger a release.

Changelog preview (truncated)

_{Preview via python-semantic-release and conventional commits.}

[github-actions]

📊 Benchmark Results

Click to see benchmark comparison

📊 Benchmark Comparison (threshold: 1.3x)
============================================================

➖ Unchanged (within threshold):
  ➖ test_simple_dict_document_store: 0.0019s → 0.0019s (-0.3%)
  ➖ test_sqlite_document_store: 0.0021s → 0.0021s (-0.1%)
  ➖ test_local_sparql_store: 0.0371s → 0.0378s (+1.8%)
  ➖ test_oneof_subschema: 0.0602s → 0.0599s (-0.4%)
  ➖ test_enum_docstrings: 0.0500s → 0.0539s (+7.9%)
  ➖ test_subclass_inheritance: 0.0535s → 0.0547s (+2.2%)
  ➖ test_class_hierarchy: 0.0508s → 0.0526s (+3.6%)
  ➖ test_core[v1]: 0.0352s → 0.0359s (+2.2%)
  ➖ test_core[v2]: 0.0433s → 0.0439s (+1.3%)
  ➖ test_schema_generation[v1]: 0.0017s → 0.0017s (-1.0%)
  ➖ test_schema_generation[v2]: 0.0037s → 0.0039s (+6.5%)
  ➖ test_simple_json: 0.0007s → 0.0007s (+0.1%)
  ➖ test_complex_graph: 0.0016s → 0.0016s (+0.2%)

============================================================
Summary: 0 regressions, 0 improvements, 13 unchanged
============================================================

✅ No significant performance regressions

Threshold: 1.3x (30% slower triggers a regression warning)

Note: Benchmarks are informational only and won't fail the build.

_{💡 Tip: Download the benchmark-results artifact for detailed JSON data}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[simontaurus]

Not reverting. Reverting to required now and later adopting a third shape would be two breaking changes where one has already been paid for.

#159 stays open: the counter-proposal and the chain-resolution question it raises are still unanswered, and whatever settles them can land as one change from here.