The message `JOINING all threads` is unfortunately not deterministic, it can happen that it just is not printed in the logs.
For Monerod container the message is set to `RPC server started ok`. This message appears in both a test run that was hanging with `JOINING all threads` and a successful run. Initially the message was set to `core RPC server started ok` with `core` being a variable value. We assume that `core` does not change, but did not to further code analysis what values it can be.
For Monero Wallet RPC container the message is set to `Run server thread name: RPC` which is what it was set to initially. After several container runs this message seems to be reasonable - there are no recorded issues of the Wallet RPC container hanging, but we had problems with Monerod in the past.
1. Split up image::Monero into Monerod and MoneroWalletRpc
2. Don't use `bash` to run the internal command. Instead we disable
the entrypoint script as per https://github.com/XMRto/monero#raw-commands
3. Remove the start up delay by listening for the correct log message.
To make this more resilient, we make the log level NOT configurable and
instead always log verbosely.
