First got to share appreciation of node red for how easy it is to jump into and all the reasons why it's so popular. Now I would like to share some thoughts on a closely related use-case I got from a similar low-code system called Pentaho ETL (mostly dead and abandoned these days unfortunately). These work very similar in most aspects, but I'd like to share a key difference:
Node red is stream focused, where each flow runs forever, typically continuously listening for input. What about a new type of flow which has a set start (and eventually an end)? Ie. it will generate a number of input messages (for example 1), but then that message will go through all the nodes (possibly generating more messages along the way), but eventually it all completes and the flow exits/stops.
Previously I'd schedule this kind of program from cron, but of course it could be scheduled from another flow within node red itself.
Main benefit is you can always know what the last message is. Take for example a flow which injects a message once per day, then that message triggers delete job to delete old log files. Eventually I get one message per file. I can write to log when it starts and each log file is deleted. But can't log when it ends (at least not in any feasible way).
With this new type of flow with a defined start and end, it is also a lot easier to aggregate / group messages on certain fields to calculate sum/max/min/avg etc. Node red can do something similar, but much more difficult since the system is designed completely different, running forever.
Not sure if this made sense or not? Just thinking outside the box here, perhaps out of scope for node red, but it is the biggest feature I'm missing.