pyln-testing: use files for stdout and stderr, not threads.

Some flakes are caused by weird races in this code.  Plus, if we
get things to write straight to files, we might see things in
there on post-mortem which happen after the python runner exits.

It's a bit less efficient, but much simpler.  Let's see if it helps!

Some tests need a rework now, since we don't get a failure (except
eventual timeout), but they're simpler.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

tests/test_misc.py

@@ -109,13 +109,13 @@ def test_bitcoin_failure(node_factory, bitcoind):
     # Ignore BROKEN log message about blocksonly mode.
     l2 = node_factory.get_node(start=False, expect_fail=True,
                                allow_broken_log=True)
-    with pytest.raises(ValueError):
-        l2.start(stderr=subprocess.PIPE)
+    l2.daemon.start(wait_for_initialized=False)
+    # Will exit with failure code.
+    assert l2.daemon.wait() == 1
     assert l2.daemon.is_in_stderr(r".*deactivating transaction relay is not"
-                                  " supported.") is not None
-    # wait_for_log gets upset since daemon is not running.
-    wait_for(lambda: l2.daemon.is_in_log('deactivating transaction'
-                                         ' relay is not supported'))
+                                  " supported.")
+    assert l2.daemon.is_in_log('deactivating transaction'
+                               ' relay is not supported')
 
 
 def test_bitcoin_ibd(node_factory, bitcoind):
@@ -1207,8 +1207,10 @@ def test_rescan(node_factory, bitcoind):
     l1.daemon.opts['rescan'] = -500000
     l1.stop()
     bitcoind.generate_block(4)
-    with pytest.raises(ValueError):
-        l1.start()
+    l1.daemon.start(wait_for_initialized=False)
+    # Will exit with failure code.
+    assert l1.daemon.wait() == 1
+    assert l1.daemon.is_in_stderr(r"bitcoind has gone backwards from 500000 to 105 blocks!")
 
     # Restarting with future absolute blockheight is fine if we can find it.
     l1.daemon.opts['rescan'] = -105
@@ -1236,14 +1238,19 @@ def test_bitcoind_goes_backwards(node_factory, bitcoind):
     bitcoind.start()
 
     # Will simply refuse to start.
-    with pytest.raises(ValueError):
-        l1.start()
+    l1.daemon.start(wait_for_initialized=False)
+    # Will exit with failure code.
+    assert l1.daemon.wait() == 1
+    assert l1.daemon.is_in_stderr('bitcoind has gone backwards')
 
     # Nor will it start with if we ask for a reindex of fewer blocks.
     l1.daemon.opts['rescan'] = 3
 
-    with pytest.raises(ValueError):
-        l1.start()
+    # Will simply refuse to start.
+    l1.daemon.start(wait_for_initialized=False)
+    # Will exit with failure code.
+    assert l1.daemon.wait() == 1
+    assert l1.daemon.is_in_stderr('bitcoind has gone backwards')
 
     # This will force it, however.
     l1.daemon.opts['rescan'] = -100
@@ -1690,7 +1697,7 @@ def test_newaddr(node_factory, chainparams):
     assert both['bech32'].startswith(chainparams['bip173_prefix'])
 
 
-def test_bitcoind_fail_first(node_factory, bitcoind, executor):
+def test_bitcoind_fail_first(node_factory, bitcoind):
     """Make sure we handle spurious bitcoin-cli failures during startup
 
     See [#2687](https://github.com/ElementsProject/lightning/issues/2687) for
@@ -1699,7 +1706,9 @@ def test_bitcoind_fail_first(node_factory, bitcoind, executor):
     """
     # Do not start the lightning node since we need to instrument bitcoind
     # first.
-    l1 = node_factory.get_node(start=False)
+    l1 = node_factory.get_node(start=False,
+                               allow_broken_log=True,
+                               may_fail=True)
 
     # Instrument bitcoind to fail some queries first.
     def mock_fail(*args):
@@ -1708,22 +1717,17 @@ def test_bitcoind_fail_first(node_factory, bitcoind, executor):
     l1.daemon.rpcproxy.mock_rpc('getblockhash', mock_fail)
     l1.daemon.rpcproxy.mock_rpc('estimatesmartfee', mock_fail)
 
-    f = executor.submit(l1.start)
-
-    wait_for(lambda: l1.daemon.running)
-    # Make sure it fails on the first `getblock` call (need to use `is_in_log`
-    # since the `wait_for_log` in `start` sets the offset)
-    wait_for(lambda: l1.daemon.is_in_log(
-        r'getblockhash [a-z0-9]* exited with status 1'))
-    wait_for(lambda: l1.daemon.is_in_log(
-        r'Unable to estimate opening fees'))
+    l1.daemon.start(wait_for_initialized=False)
+    l1.daemon.wait_for_logs([r'getblockhash [a-z0-9]* exited with status 1',
+                             r'Unable to estimate opening fees',
+                             r'BROKEN.*we have been retrying command for --bitcoin-retry-timeout=60 seconds'])
+    # Will exit with failure code.
+    assert l1.daemon.wait() == 1
 
     # Now unset the mock, so calls go through again
     l1.daemon.rpcproxy.mock_rpc('getblockhash', None)
     l1.daemon.rpcproxy.mock_rpc('estimatesmartfee', None)
 
-    f.result()
-
 
 @pytest.mark.developer("needs --dev-force-bip32-seed")
 @unittest.skipIf(TEST_NETWORK != 'regtest', "Addresses are network specific")
@@ -2077,8 +2081,9 @@ def test_new_node_is_mainnet(node_factory):
     del l1.daemon.opts['network']
 
     # Wrong chain, will fail to start, but that's OK.
-    with pytest.raises(ValueError):
-        l1.start()
+    l1.daemon.start(wait_for_initialized=False)
+    # Will exit with failure code.
+    assert l1.daemon.wait() == 1
 
     # Should create these
     assert os.path.isfile(os.path.join(netdir, "hsm_secret"))