We have a log statement for when connection establishment fails, but it mistakenly is logged regardless of whether the establishment actually failed. We should add a check that the status is non-OK before logging.

It's only logged at a very high diagnostic level, so this shouldn't generally affect production clusters.

https://github.com/mongodb/mongo/blob/ac1d0a40505966c5a1a83e97f5cffc0c6efd7578/src/mongo/executor/connection_pool.cpp#L1222-L1228