Merge #585: bitcoind: call wait in drop to reap process

c08ea32 bitcoind: call wait in drop to reap process (Nick Johnson)

Pull request description:

  Over in [rust-psbt we are seeing integration test failures](https://git.rust-bitcoin.org/rust-bitcoin/rust-psbt/actions/runs/58/jobs/8/attempt/1) which result in a CI runner spinning for hours, even though the test failed in a minute. Setting aside why the test failed (looks bitcoind related but dunno what), I am wondering if we should have a stronger guarantee in the drop to clean up the process. I looked back in the [old archive for the original impl reasoning](rust-bitcoin/bitcoind#52) and don't see much.

  I could also see this being a big no-no in a Drop impl, but for a high level thing like the bitcoind process maybe better than not?

ACKs for top commit:
  tcharding:
    ACK c08ea32

Tree-SHA512: 9a1ff79f4f8ad3735363cc385f235fd3dc99643965b4f134de5514014167f8e41baf313384173d8404e650d2cfcf74d1f99dcd3edf4f596a5922a49cf83a6d5a

Author: tcharding

bitcoind/src/lib.rs

@@ -573,10 +573,13 @@ impl BitcoinD {
 
 impl Drop for BitcoinD {
     fn drop(&mut self) {
+        // Frist attempt graceful shutdown for persistent directories,
+        // always fallback to force kill and wait for process to be reaped.
         if let DataDir::Persistent(_) = self.work_dir {
             let _ = self.stop();
         }
         let _ = self.process.kill();
+        let _ = self.process.wait();
     }
 }