Fix 0.6.2 compatibility test shutdown

[tnull]

Stop and drop the old compatibility node from a blocking region.

This avoids a Tokio runtime-drop panic in the async test.

Honestly not quite sure what changed that this only surfaces now that CI is back (likely an unrelated refactoring while CI was gone is to blame). But we need to fix it so CI finally is green again.

[ldk-reviews-bot]

👋 Thanks for assigning @joostjager as a reviewer!
I'll wait for their review and will help manage the review process.
Once they submit their review, I'll check if a second reviewer would be helpful.

[joostjager]

AI analysis regarding why now:

What is happening: v0.6.2's Node::start() creates and stores its own tokio::runtime::Runtime. In the backwards-compatibility test we call node_old.stop() from inside #[tokio::test], so shutdown runs while the test is already executing in a Tokio async context.

The subtle part is the tokio_unstable cfg. In v0.6.2, Node::stop() clones the runtime into metrics_runtime under #[cfg(tokio_unstable)] to log active task metrics. The shutdown body uses block_in_place around its internal runtime.block_on(...) calls, but that extra clone remains alive until the end of stop(). When stop() returns to the async test, the final runtime drop can happen in a context where Tokio does not allow blocking, and Tokio panics with:

Cannot drop a runtime in a context where blocking is not allowed. This happens when a runtime is dropped from within an asynchronous context.

[joostjager]

The vss_no_auth integration test uses spawn_blocking for a similar code block.