Add throttle option for logging

[HannesBachter]

adds the option to use Logger.log*_throttle

[pschillinger]

Thanks for adding! Would you see a way that entries can also be removed again eventually so that this dict can't grow infinitely?

[HannesBachter]

The update function would only update the entry within the dict (whose key is a combination of the logger level and the text) - so the dictionary would only have the max length of the number of Logger.log*_throttles
This idea works only until the text within the log is not static, and in the worst case changing for every log (e.g. a timestamp).
As a safety-net I could add the clearing of the dict from all entries that are older than e.g. one hour (as the entries are {log_id: timestamp})?

[pschillinger]

I think it would be sufficient as a safety-net to just check whether the number of dict entries exceeds a certain limit and then remove entries. As long as the limit is not exceeded, it's probably fine to keep old entries.
As a further idea, I would suggest to log a warning message if the size limit is reached because, as you pointed out, this would likely be caused by improper usage of the throttle such as a non-static messages.

[dcconner]

Would it be reasonable to include a limit on dictionary size?
https://itecnote.com/tecnote/python-how-to-limit-the-size-of-a-dictionary/

[HannesBachter]

@pschillinger ping :)

[dcconner]

@HannesBachter , with @pschillinger comments incorporated, would you be willing to submit a PR against the Noetic branch at https://github.com/FlexBE/flexbe_behavior_engine . That's where we are maintaining the behavior engine going forward. I'll take care of getting in to the ROS 2 branches. I don't plan to make changes to this old version.

[HannesBachter]

Opened a new PR here: FlexBE/flexbe_behavior_engine#14