Enable improved panic error reporting in runtime

[hitchho]

Enables the sp-io/improved_panic_error_reporting feature for the hydradx runtime, so runtime panics reach RPC callers as readable messages instead of opaque wasm traps.

Motivation

While debugging #1491, a parameter-decode failure in DryRunApi.dry_run_call surfaced over public RPC as:

Execution failed: Execution aborted due to trap: wasm trap: wasm `unreachable` instruction executed

The actual panic message — Bad input data provided to dry_run_call: Codec error. Nothing bad happened: someone sent an invalid transaction to the node. — only exists in node logs, invisible to the RPC caller. It cost a chopsticks fork to see it, and the symptom was initially misdiagnosed as a runtime bug.

With this feature, sp-io's wasm panic handler calls the ext_panic_handler_abort_on_panic host function instead of executing unreachable, and the client executor returns Runtime panicked: <message> (sc-executor AbortedDueToPanic) — making every future decode error, defensive panic, and panic! diagnosable directly from the RPC response.

Safety

• Client executors: sc-executor-wasmtime has captured the message via HostState::panic_message since 2022; smoldot implements ext_panic_handler_abort_on_panic_version_1.
• Relay-chain PVF validation: the PVF executor's host function list omits panic_handler, but it instantiates with allow_missing_func_imports: true, so the import resolves to a stub. A panic during validation then traps exactly as it does today — outcome-identical (candidate invalid).
• The feature only affects the #[panic_handler]/#[alloc_error_handler] compiled into the wasm blob; no behavior change on the success path.

For reference, this is the same mechanism substrate/client/executor/runtime-test uses upstream.