currently we use AsyncMulticaster to multicast the search index command on all hosts. The  AsyncMulticaster is intended to send out commands without providing a mechanism for waiting on their responses. This means we can't block the main thread from continuing to execute before the search index commands have finished on all mongods/mongots.

To be able to block, we need to instead loop over all of the hosts and one by one scheduleRemoteCommand where the command is the userCmd (create/drop/update searchindex command) wrapped in a _shardsvrRunSearchIndexCommand. 

something like this:

            executor::RemoteCommandResponse response(
                host, Status(ErrorCodes::InternalError, "Internal error running command"));

            auto callbackHandle = uassertStatusOK(executor->scheduleRemoteCommand(
                request,
                [&response](const executor::TaskExecutor::RemoteCommandCallbackArgs& args) {
                    response = args.response;
                }));

            // Block until the command is carried out
            executor->wait(callbackHandle);