Complete the OverPy lexical and grammar surface

[Teakowa]

Parent: #88

Goal

Complete the pinned OverPy source grammar as one audited language-syntax capability rather than continuing to fix parser residuals one construct at a time.

Context

Earlier syntax work established a broad baseline, but later real-project corpus runs still exposed lexical and grammar forms that were absent from the original inventory. Corpus failures are useful evidence, but they must not define the roadmap unit.

This workstream is part of the source-language compatibility program in #88. The implementation should be driven by a complete pinned-upstream syntax inventory and validated by conformance plus real-project evidence.

Scope

• Audit the pinned OverPy lexer/parser surface as a whole, including literals, strings, line continuation, multiline and parenthesized expressions, declarations, statements, control flow, collection syntax, comprehensions/lambdas, annotations, and other grammar forms accepted by the compatibility target.
• Close all demonstrated opy-rs lexer/parser gaps found by that audit within the declared compatibility surface.
• Preserve source spans, trivia/provenance needed by diagnostics and tooling.
• Keep invalid/reference-rejected forms as structured source diagnostics.
• Use the real-project corpus to detect inventory omissions after the audited syntax surface is implemented.

Non-goals

• Preprocessor, macro, or #! directive semantics; those belong to the dedicated preprocessing workstream.
• Builtin/action/value/member availability or canonical Workshop catalog ownership.
• Compiler-output identity, optimizer behavior, or Workshop -> OPY reconstruction.
• Project-specific parser workarounds.

Acceptance criteria

• The pinned OverPy syntax inventory has executable positive and negative coverage at the appropriate structural level.
• Every accepted syntax form in the declared inventory reaches the OPY semantic pipeline with stable provenance.
• Known real-project lexical/parser residuals are either resolved or shown by independent evidence to be outside the declared compatibility target.
• A full corpus rerun does not expose an unowned lexer/parser grammar form that belongs to this workstream.
• No canonical Workshop semantics are duplicated in opy-rs.

PR budget

Target: 1 implementation PR. Maximum: 2 non-release PRs. Split only for reviewable code boundaries, not by individual syntax construct.

[Teakowa]

Reopened after the #142 full-corpus rerun exposed a remaining OPY-owned lexical boundary in real-world/ow1-emulator: lexer line continuation (lex-error). PR #146 completed the audited statement grammar slice, but this residual is still inside #141's declared lexer/grammar scope and therefore the category acceptance criteria are not yet satisfied.

Keep this under #141 rather than creating a new micro issue. The workstream still has its allowed second non-release implementation PR available. #143 may continue in parallel; this correction does not change the category-first roadmap or the overall 10-PR budget.