P6.x: convert emitter::block_uses_emit to read the checker's own resolved Callee

[accuser]

Summary

• Scope: emitter::block_uses_emit — the per-handler twin deciding emit_service/emit_agent's
  own deps.__eventsDispatch parameter threading — now reads the checker's own resolved
  Callee::Capability{cap:"Events",op:"emit"} instead of matching a bare Ident("Events") receiver by
  name.
• Addresses: the real inconsistency P6.x: thread per-unit Callee data forward; convert unit_table_uses_emit and called_cross_context_services #1202's own review found and documented in a code comment,
  not left silent.
• Realises: the two Events-detection checks (unit_table_uses_emit, project-wide, and
  block_uses_emit, per-handler) now fully agree on every input.

The inconsistency this closes

Before #1202, both unit_table_uses_emit (project.rs) and block_uses_emit (emitter.rs) matched a
bare Ident("Events") receiver by name — the same syntactic approach, both accepted as a deliberate
"approximation" per block_uses_emit's own prior doc comment (mirroring block_uses_send's sibling
precedent). #1202 converted unit_table_uses_emit to read the checker's own resolved Callee, but left
block_uses_emit untouched — its own review caught this and documented it explicitly rather than
papering over it with a fixture that would have failed tsc --strict.

A locally-declared type also named Events with its own static emit method (legal Bynk, resolves to
Callee::Static, not Callee::Capability) makes the two disagree: unit_table_uses_emit correctly
skips compose/fan-out generation, while block_uses_emit still threads a deps.__eventsDispatch
parameter with nothing left to supply it — a real tsc type error, not just cosmetic drift.

Why this was mechanical, unlike #1202's own project.rs plumbing

Every one of block_uses_emit's callers already had a TypedCommons/CheckedProgram in scope —
emit_service/emit_agent take commons, the ir::lower callers take program. So this is purely a
signature change (block_uses_emit(b: &Block, callees: &HashMap<ExprId, Callee>) -> bool) plus
call-site threading — no new data-flow plumbing needed, unlike #1202's own RunChecks::Checked
threading. block_uses_send needs no matching conversion: a ~> send is a real Statement::Send AST
variant, not a method call that could be shadowed, so it was never approximate the way
block_uses_emit's method-name match was.

Correctness evidence

• BYNK_BLESS=1 cargo test -p bynkc --test e2e bless_positive_fixtures touched zero existing files.
• 1204_events_emit_shadowed_by_local_type — the exact shape that failed tsc --strict under
  P6.x: thread per-unit Callee data forward; convert unit_table_uses_emit and called_cross_context_services #1202 alone (found during that PR's own review, documented, deliberately not landed broken) — now
  compiles and passes tsc --strict clean. cargo test -p bynkc --test tsc_verify (real tsc --strict
  over every fixture's emitted output) is clean, including this one.
• ast_importers unaffected (still 8).

cargo build --workspace / cargo clippy --workspace --all-targets / cargo fmt --all -- --check /
cargo test --workspace all clean (164 test-result blocks, 0 failures, including
emitted_typescript_passes_tsc_strict).

Done when

• block_uses_emit reads Callee, not a bare-Ident name match.
• All 10 call sites converted (emitter.rs, emitter/emit.rs ×7, ir/lower.rs ×3).
• The fixture that exposed the disagreement now compiles and passes tsc --strict.
• unit_table_uses_emit's own doc comment updated — the "known follow-on" it named in P6.x: thread per-unit Callee data forward; convert unit_table_uses_emit and called_cross_context_services #1202 is
  resolved, not just left stale.
• block_uses_send confirmed not needing the same fix, with the reasoning stated.
• ast_importers unaffected (still 8) — stated explicitly.
• design/pending/p6-block-uses-emit-callee.md added (patch level — internal only, no language
  surface change beyond fixing a real tsc-failing edge case).
• A new ADR records this decision if the reviewing bot calls for one; its number is assigned at merge.
  Part of Phase 6 — The IR (spine) #1137 — a further step toward project.rs's own slice 6 cleanup.