feat(opy): complete bounded canonical lowering

[e54-bot]

Summary

• complete bounded OPY-to-canonical-WIR lowering for literal strings, f/w/b/c modifiers, literal dictionary lookup, contextual chaseAtRate, and null defaults
• preserve canonical text contexts for debug, print, and HUD helpers, including constant folding for literal format arguments
• add oracle-backed Complete OPY-to-canonical-WIR lowering for the supported source language #145 lowering tests and refresh compiler/support evidence; retain nested indexed writes and unrelated builtin/parser gaps as explicit owner boundaries

Verification

• cargo fmt --all -- --check
• python3 -m unittest discover -s compatibility/tests
• cargo clippy --workspace --all-targets --all-features -- -D warnings
• cargo test --workspace --all-targets --all-features
• full 61-fixture corpus: 61/61 compile-status matches, 40 matches, 19 declared known gaps, and 2 declared unsupported boundaries

Scope note

The remaining chase fixture semantic gap is tracked by opy-rs#144; the broad issue-28 oracle snapshot remains a declared #88 corpus boundary because the released Workshop parser cannot currently reparse its uppercase hexadecimal spelling. No unowned OPY lowering gap is introduced by this PR.

Fixes #145

[Teakowa]

Major — crates/opy-rs/src/compiler/mod.rs:3364

The new literal-dict fast path only handles a matching key. A valid literal miss (for example {"a": 1}["b"]) falls through to generic Expr::Dict lowering and is rejected. The pinned OverPy implementation instead lowers dict access through key/value arrays; for an all-literal miss .index folds to -1, then __valueInArray__ folds the negative index to null. Handle the missing-key case and add oracle-backed coverage before claiming compilation/opy-literal-dict-lookup end-to-end supported.

Major — compatibility/support-matrix.json / Fixes #145

#145 explicitly subsumed #60, and #60 was closed as “superseded by #145”, but this PR still leaves deeper indexed writes “tracked by #60” while closing #145. The current compiler expectations also retain synthetic/issue-31-positive as a known-gap owned by opy-rs#145. Merging this as Fixes #145 therefore leaves declared lowering gaps without a live follow-up owner and does not satisfy #145’s category-level completion contract. Either complete/reclassify those #145-owned boundaries here, or keep #145 open (within its two-PR budget) and give the remaining gap an active owner.