cargo-green emitter fixes: M: Clone bound on bare-generic returns (E0599) + Optional receiver unwrap before method call (E0624)

[gunbai-bot]

Route-A last mile: fix the two emitter root causes that blocked cargo build on the faithful emitted crate, reducing the error count from 5 → 2 (the remaining 2 are E0091 Compose phantom-type-param, HOLD per operator ruling).

What and why

E0599 (std_measure.rs:126): emit_fn_def called emit_type_params with no bounds even when the function's return type is a bare generic param (e.g. M in fn measure_count<Q,S,M>(...) -> M). The emitter accesses the field via m.count.clone() which requires M: Clone (because count: Box<M> and Box<M>: Clone requires M: Clone). Fix: detect authored_name_at(inferred) ∈ generic_param_names in emit_fn_def and switch to emit_type_params_with_clone_bound for that param.

E0624 (std_realization_schedule.rs:204/228): method calls on Optional receivers were emitted without unwrapping. When batch0 = schedule.first() (returning Optional<List<T>>), calling batch0.count() emitted batch0.len() — but Option<Vec<T>> has no .len(). Fix: detect resolved_type(receiver).return_cardinality == CardOptional in the AlgebraMethodSemantics else-branch and append .expect("fail-closed: ...") before template expansion. Reuses the same unwrap pattern as rust_call_arg_fail_closed_unwrap for function arguments.

Both fixes land in .dag (the authority: src/v1/05_emit_rust.dag) and are hand-synced to the seed (src/v1/stage0/src/v1_compiler_emit_rust.rs). No seed-only changes.

Evidence

--emit-fresh receipt showing 5 → 2 errors after fixes:

$ ./target/debug/regen_stage0 --emit-fresh <dir>
$ cargo build   # in <dir>
error[E0091]: type parameter `MachineConstraint` is never used   ← HOLD (Compose)
error[E0091]: type parameter `Algebra` is never used             ← HOLD (Compose)

E0599 × 1 and E0624 × 2 are gone.

Test plan

cargo test -p v1-compiler-tests generic_return_clone_bound   # 2/2 pass
cargo test -p v1-compiler-tests optional_receiver_method_unwrap  # 2/2 pass
cargo fmt --all --check   # clean

Each test has a positive gate (the fix fires) and a negative control (the gate is not too broad).