fix: bump feerate and tx broadcast timeouts to 15s

The 5-second node-level timeout fires before Electrum can complete
a single request (10s timeout), causing FeerateEstimationUpdateTimeout
on node start. Bump both FEE_RATE_CACHE_UPDATE_TIMEOUT_SECS and
TX_BROADCAST_TIMEOUT_SECS from 5s to 15s.

Upstream fixed this in lightningdevkit/ldk-node@40c4b09 (bumped to
10s). We use 15s for extra margin with slow Electrum servers.

Author: ovitrif

src/config.rs

@@ -85,10 +85,10 @@ pub(crate) const LDK_EVENT_HANDLER_SHUTDOWN_TIMEOUT_SECS: u64 = 30;
 pub(crate) const BACKGROUND_TASK_SHUTDOWN_TIMEOUT_SECS: u64 = 5;
 
 // The timeout after which we abort a fee rate cache update operation.
-pub(crate) const FEE_RATE_CACHE_UPDATE_TIMEOUT_SECS: u64 = 5;
+pub(crate) const FEE_RATE_CACHE_UPDATE_TIMEOUT_SECS: u64 = 15;
 
 // The timeout after which we abort a transaction broadcast operation.
-pub(crate) const TX_BROADCAST_TIMEOUT_SECS: u64 = 5;
+pub(crate) const TX_BROADCAST_TIMEOUT_SECS: u64 = 15;
 
 // The timeout after which we abort a RGS sync operation.
 pub(crate) const RGS_SYNC_TIMEOUT_SECS: u64 = 5;