fix(spec): deduplicate derived completers

[jdx]

Emits each identical built-in completion node only once per command. Multiple fields sharing a completion name and type (for example path-valued flags with value_name = "PATH") no longer produce repeated complete "path" nodes. Distinct names or types retain declaration order.

The parser and facade test suites pass, including a regression that parses both fields and asserts one emitted completer. PLAN now records this portion of the broader canonical-KDL gap as complete.

Stacked on #1071.

---

Note

Low Risk
Cold-path spec emission and tests only; parsing and binding behavior are unchanged.

Overview
Derived KDL emission now writes each identical built-in complete node only once per command. When several flags or arguments share the same completion name and type (for example two value_name = "PATH" fields with a path hint), write_completion_types tracks what was already emitted instead of repeating complete "path" type="path".

A facade regression test (CompletionDedup) parses both flags and asserts a single matching completer line in to_kdl(). PLAN.md records that this slice of the broader canonical-KDL work is done; full parse/serialize canonicalization is still open.

^{Reviewed by Cursor Bugbot for commit cda8d65. Bugbot is set up for automated code reviews on this repo. Configure here.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Central YAML (base), Organization UI (inherited)

Review profile: CHILL

Plan: Pro Plus

Run ID: d8725d38-a338-43d6-97ab-3edfb7b16b76

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[github-actions]

Instruction counts

benchmark	trend	instructions	Δ	wall (min)	Δ
markdown	▁█▇█▇▇▇▇▇	198,757,447 → 198,508,246	-0.13%	18.56 → 18.93ms	+2.00%
startup	▁████████	1,214,963 → 1,214,976	+0.00%	1.36 → 1.46ms	+7.62%

No instruction-count regression above 1%.

_{Only instruction counts gate. Wall clock is shown for context — on identical hardware it moves 4-20% run to run.}

_{Measured by tak — instruction-counted CLI benchmarks, stored in this repository's git notes.}

Shadow comparison

Parsing mise use -g node@20 against a shadow of mise's committed spec.
Reported, not gated: the shadow grows as the derive learns to express more, so
what to watch is the ratio rather than either column.

framework	instructions, cold parse	vs usage
usage	4253	—
argh	6292	1.5x
clap	5893640	1385x
bpaf	21917948	5153x

                                              min       p01       p10    median
usage-rs: argv -> struct                      195       198       202       210  ns
argh: argv -> struct                          269       274       282       297  ns
clap: build tree + parse -> struct         481454    482899    489073    500192  ns
bpaf: build parser + parse -> struct      1644794   1644794   1654953   1682691  ns

usage: argv -> struct                             209 ns      0.21 µs
clap: build tree + parse -> struct             491458 ns    491.46 µs
clap: parse -> struct, tree reused              23599 ns     23.60 µs
clap: build tree only                          307962 ns    307.96 µs

_{cda8d65d4b65 vs 0b137759b66b · measured on the runner, not pushed to the history.}