Split test_general into stable and experimental targets

[kraenhansen]

Problem

The upstream test_general test in the Node.js repository compiles a single target with NAPI_EXPERIMENTAL, which links against all experimental Node-API symbols (node_api_set_prototype, node_api_post_finalizer). This means the addon cannot be loaded on runtimes that don't export every experimental symbol.

In the CTS, this forces all test_general JS test files to guard loadAddon behind a check for every experimental feature the addon links against. The result is that even stable API tests (like napi_strict_equals, napi_typeof, napi_instanceof, etc.) are silently skipped on runtimes that don't support all experimental features.

Proposed solution

Split the upstream test_general into separate targets:

1. Stable target — compiles without NAPI_EXPERIMENTAL, includes all stable API functions
2. Experimental target(s) — one per experimental feature, compiled with the appropriate NAPI_EXPERIMENTAL define

This would allow the CTS to test stable APIs independently of experimental feature support.

Current workaround

The CTS ports test_general as a single experimental addon (matching upstream), with all JS tests guarded behind experimentalFeatures.setPrototype && experimentalFeatures.postFinalizer.

References

• Upstream source: https://github.com/nodejs/node/tree/main/test/js-native-api/test_general
• Upstream binding.gyp defines NAPI_EXPERIMENTAL on the single target