Improves reconnection responsiveness.

As the user points out in

   #205

wait_for_all might be the cause of delay in the
reconnection. I could not observe any improvement in my
local setup but wait_for_one_error look the correct token
and all tests pass so I am willing to change it.

Author: mzimbres

README.md

@@ -676,6 +676,12 @@ https://lists.boost.org/Archives/boost/2023/01/253944.php.
 
 ## Changelog
 
+### Boost 1.87
+
+* ([Issue 205](https://github.com/boostorg/redis/issues/205))
+  Improves reaction time to disconnection by using `wait_for_one_error`
+  instead of `wait_for_all`.
+
 ### Boost 1.85
 
 * ([Issue 170](https://github.com/boostorg/redis/issues/170))

include/boost/redis/detail/runner.hpp

@@ -93,7 +93,7 @@ class runner_op {
             [this](auto token) { return runner_->health_checker_.async_check_health(*conn_, logger_, token); },
             [this](auto token) { return runner_->async_hello(*conn_, logger_, token); }
          ).async_wait(
-            asio::experimental::wait_for_all(),
+            asio::experimental::wait_for_one_error(),
             std::move(self));
 
          logger_.on_runner(ec0, ec1, ec2);

test/test_conn_exec.cpp

@@ -152,6 +152,9 @@ BOOST_AUTO_TEST_CASE(correct_database)
    auto const pos = value.find("db=");
    auto const index_str = value.substr(pos + 3, 1);
    auto const index = std::stoi(index_str);
+
+   // This check might fail if more than one client is connected to
+   // redis when the CLIENT LIST command is run. 
    BOOST_CHECK_EQUAL(cfg.database_index.value(), index);
 }