Record the current-directory and directory-target product decision in ADR history

[Teakowa]

Goal

Record the durable product/integration decision behind first-class current-directory and directory project targets without rewriting the earlier explicit-entry decision history.

Context

Issue #317 and PR #318 established a material extension to Wright's input and source-owner integration contract:

• omitted INPUT selects the current working directory instead of stdin;
• - remains explicit stdin;
• directories are first-class project targets;
• Wright performs only the minimum inspection needed to identify a source owner and must fail explicitly on ambiguous ownership;
• effective entry, project root, and source closure remain owned by the source implementation;
• provider-backed directory targets use the owner/LPP boundary rather than a Wright project loader.

The current CLI/driver documentation describes this behavior, but the ADR history currently stops at ADR-0013's earlier entry-based source-provider seam. The rationale for extending that seam to directory targets should remain recoverable independently from current implementation and Issue state.

Scope

• Classify Make current-directory targets first-class and delegate automatic entry discovery to source owners #317/feat(input): support current-directory project targets #318 against ADR-0013 and the current architecture/CLI contracts.
• Add the smallest new Wright ADR needed to preserve the rationale for current-directory defaults, first-class directory targets, minimal source-owner detection, explicit ambiguity, and delegation of project semantics to source owners.
• Preserve ADR-0013 as historical decision context rather than rewriting it to imply that directory targets were part of the earlier decision.
• Link externally owned LPP wire rationale instead of duplicating protocol authority in Wright.
• Reconcile current architecture/CLI links only where needed to make the decision history discoverable.

Non-goals

• Changing the implemented input semantics from feat(input): support current-directory project targets #318.
• Adding a Wright workspace/project graph, project manifest, recursive source loader, or source-language precedence table.
• Reimplementing OPY/DEL project discovery in Wright.
• Recording current provider versions, release state, or real-project migration progress in ADR text.
• Duplicating LPP wire-contract rationale owned by language-provider-protocol.

Acceptance criteria

• Wright's ADR history explains why omitted input targets the current directory while - remains explicit stdin.
• The rationale for treating a directory as a source-owner target rather than a Wright-loaded source closure is recoverable.
• Explicit ambiguity/no-hidden-precedence behavior is represented as a durable product/integration consequence where warranted.
• ADR-0013 remains historically accurate and links cleanly to the later directory-target decision.
• Current CLI/architecture docs remain the authority for present behavior; the ADR records rationale/history only.
• Cross-repository protocol rationale is linked to the LPP owner decision rather than copied into Wright.

Ownership

Wright owns CLI/default-target UX, source-owner selection, ambiguity handling, and product orchestration. OPY/DEL project semantics remain source-owned; LPP owns the process/wire target handoff.