Bug: MongoDB container is reported ready before the server accepts connections

Look at the MongoDbContainer class in testcontainers/mongodb.py to understand the current wait strategy. The fix involves updating the wait strategy to check for the 'Listening on' log entry with address '0.0.0.0' in JSON logs. Run the provided reproduction script to verify the bug and test the fix. Ensure the change works for MongoDB 4.4 and newer versions.

Describe the bug

When a root user is configured, which MongoDbContainer always does, the mongo image starts twice. A temporary server listening only on localhost creates the user and exits, then the real server starts. Both log "Waiting for connections", and the container's wait strategy matches the first one. So start() returns while no server is reachable from outside the container, and on a slow host the client times out before the real server is up.

I expected start() to return once the real server accepts connections. A line only the real server logs is the "Listening on" entry with address 0.0.0.0 (JSON logs, MongoDB 4.4 and newer), and waiting for r'"Listening on".*"address":"0\.0\.0\.0"' makes the snippet below pass.

To Reproduce

from testcontainers.mongodb import MongoDbContainer

# Limit the container to a tenth of a CPU, as a busy CI runner would.
with MongoDbContainer("mongo:7").with_kwargs(nano_cpus=100_000_000) as mongo:
    client = mongo.get_connection_client()
    print(client.admin.command("ping"))

pymongo.errors.ServerSelectionTimeoutError: localhost:50264: [Errno 54] Connection reset by peer (configured timeouts: socketTimeoutMS: 20000.0ms, connectTimeoutMS: 20000.0ms), Timeout: 30s

It fails every time at this limit and passes with nano_cpus=1_000_000_000.

Runtime environment

macOS 26 (arm64), Docker 29.7.2, Python 3.14.0, testcontainers 4.15.0. The wait strategy on main is the same.