feat: add calendar spreads to MultiLegStrategy

[sunli829]

Follows longbridge/developers#1251, which documents two new values on the multi-leg strategy enum.

MultiLegStrategy gains:

• CalendarCallSpread — wire value 7
• CalendarPutSpread — wire value 8

Calendar (horizontal) spreads. Accepted by TradeContext.submit_multileg, and reported back on the multi_leg field of today_orders / history_orders / order_detail and the order-changed push.

Both variants are appended after Strangle, so the discriminants of the existing variants are unchanged — this is additive, not a breaking change. Without them the SDK parsed the two new server values as Unknown and had no way to submit either strategy.

Layers touched

Layer	Files
Rust core	rust/src/trade/types.rs (with the #[strum(to_string = …, serialize = "7"/"8")] mapping)
C	c/src/trade_context/enum_types.rs
C++	cpp/include/types.hpp, cpp/src/convert.hpp (both convert() directions)
Java	java/src/types/enum_types.rs, java/javasrc/…/trade/MultiLegStrategy.java
Node.js	nodejs/src/trade/types.rs
Python	python/src/trade/types.rs, python/pysrc/longbridge/openapi.pyi
Generated	c/csrc/include/longbridge.h, nodejs/index.d.ts

Verification

• cargo clippy --all --all-features and cargo +nightly fmt --all — clean (only the pre-existing longbridge-nodejs warnings)
• cargo build -p longbridge-c / -p longbridge-java / -p longbridge-python — pass
• npm run build:debug — pass; index.d.ts regenerated
• C++: syntax-checked a TU including convert.hpp (g++ -std=c++17 -fsyntax-only); both convert() overloads resolve the new variants. Full CMake build not run.
• strum round-trip checked with a throwaway cargo test: "7"/"8" and "CalendarCallSpread"/"CalendarPutSpread" parse to the new variants, and to_string() yields the string names. The temporary test file is not part of this PR.

java/javasrc was not compiled — no JDK on this machine. The change there is two enum constants; the Rust JNI half does build.

🤖 Generated with Claude Code