fix(orchestrator): fanout succeeded inversion, memoize fee lookup, add toolchain docs

[webthreejunkie-sudo]

Summary

Closes #1336
Closes #1339
Closes #1345

---

#1336 — DX: document required Rust toolchain version

New files:

• CONTRIBUTING.md — toolchain version table (channel, components, targets), install instructions, idempotent verification steps, pre-PR checklist, and repo-specific rules
• scripts/verify_toolchain.sh — idempotent script that checks rustc stable channel, wasm32-unknown-unknown + wasm32v1-none targets, rustfmt and clippy components, and cargo check; exits 1 with remediation hints on any failure

---

#1345 — Fix: cross-contract call swallows the inner error

Root cause: execute_flow_fanout used .is_err() on all three try_* downstream calls and assigned the result to FanOutStepResult::succeeded. This inverted the semantics: a successful call set succeeded = false and a failing call set succeeded = true. all_succeeded was therefore always wrong.

Fix (orchestrator/src/lib.rs): Changed .is_err() → .is_ok() for s_ok, b_ok, i_ok.

---

#1339 — Perf: memoize the fee lookup across a batch

Root cause: execute_flow_fanout made 6 individual instance().get() storage lookups and computed splits with a hardcoded amount / 3 approximation instead of calling calculate_split. This ignored the configured split percentages entirely.

Fix (orchestrator/src/lib.rs):

• Replaced 6 individual storage reads with a single FlowRouting::from_storage(&env) call (the same pattern used by all other flow paths)
• Replaced amount / 3 with one calculate_split(&amount) cross-contract call; the result is stored in a local Vec and its four elements are reused for all three downstream steps — the fee schedule is fetched exactly once per batch invocation

Before vs after (for a 10 000-unit remittance with a 50/30/15/5 split):

	Before (hardcoded thirds)	After (memoised split)
savings	3 334	3 000
bills	3 333	1 500
insurance	3 333	500

---

Tests added (orchestrator/src/test.rs)

Test	Issue
test_fanout_all_succeed_flags_are_true	#1345 happy path
test_fanout_savings_failure_reports_succeeded_false	#1345 failure path
test_fanout_bill_failure_reports_succeeded_false	#1345 failure path
test_fanout_insurance_failure_reports_succeeded_false	#1345 failure path
test_fanout_amounts_match_calculate_split_not_hardcoded_thirds	#1339
test_fanout_rejects_zero_amount	explicit failure mode
test_fanout_rejects_short_split_vector	explicit failure mode

---

Checklist

• Change matches the summary of each issue
• Tests cover happy path + explicit failure mode for each fix
• verify_toolchain.sh is idempotent and exits non-zero on any failure
• No hard-coded limits in call sites (thresholds read from params.rs / contract storage)
• No unrelated refactors or style-only changes
• scripts/verify_toolchain.sh is executable (chmod +x)