[Dcastroro]

URL.from_text("café/path").to_iri().to_text() currently returns caf%C3%A9/path: _encode_path_parts() drops the maximal=False option for the first schemeless segment. Forward the option so textual IRI output preserves Unicode, while to_uri() still emits UTF-8 percent escapes and a colon in the first relative segment remains escaped.

Adds one regression test with three assertions for text, URI, and URI-to-IRI conversion, including a first-segment colon.

Validation on Windows CPython 3.9.25 with the repository-pinned pytest 5.2.4, Hypothesis 5.8.6 and idna 2.9:

• Regression fails on the unchanged implementation and passes after the fix.
• URL, DecodedURL and URL doctests: 108 passed. Changed-file Black 21.7b0, Flake8 3.9.2, mypy 0.910 and git diff --check pass.
• Full suite with seed 237858325326282355359246794251327781884: before 134 passed / 2 failed; after 135 passed / 1 failed. The remaining test_hostnames_ascii_nolead filtering health check also fails on the unchanged implementation and is already tracked by Tests are failing due to hypothesis filtering #186 / Avoid uppercase ASCII labels in hostname strategies #205. No health checks were suppressed.
• Full mypy reports 16 errors confined to the unchanged Hypothesis strategy/test files; both changed files pass. Other interpreter/platform combinations were not run locally.

[Copilot]

🟢 Approval recommended

The single-line forwarding of the maximal flag is a minimal, correct fix with an accompanying regression test and no outstanding objective issues.

No open issues 🎉

What changed in this PR

This PR fixes a bug in hyperlink's IRI serialization where relative (schemeless) paths lost their Unicode characters. Previously _encode_path_parts() always encoded the first schemeless segment with the default maximal=True, so textual IRI output percent-escaped Unicode (e.g. café/path → caf%C3%A9/path). The fix forwards the caller's maximal flag to _encode_schemeless_path_part(), so to_text() (maximal=False) preserves Unicode while still escaping syntactically required delimiters, and to_uri() (maximal=True) continues to emit UTF-8 percent escapes.

Changes:

• Forward the maximal option to _encode_schemeless_path_part() for the first schemeless path segment.
• Add a regression test asserting IRI text preservation, URI percent-encoding, and URI→IRI round-tripping, including a first-segment colon that must stay escaped.

File	Description
src/​hyperlink/​_url.py	Passes maximal=maximal into _encode_schemeless_path_part() so IRI text output preserves Unicode in the first relative segment.
src/​hyperlink/​test/​test_url.py	Adds test_schemeless_iri_path covering text, URI, and URI-to-IRI conversion with a first-segment colon.

_{Review effort: Balanced}