Refactor schedule to avoid duplication and be more flexible

[geonnave]

With the latest simulations, I found a new set of schedules that can be flexible in the number of devices while requiring less RAM.
This is because the current schedules are prime, to have an even frequency hopping distribution.
However, if we use 37 channels (prime number), I verified that the hopping is already pretty good at schedules for 20-40-60-80-100 nodes.

So it is possible to refactor the schedule structure like this:

1. a single pre-defined schedule, very large
2. in the beacon, instead of setting the schedule id, we set the max node size
3. the schedule in use is capped by that size

This allows much more flexible schedule adaptation, while keeping the protocol simple and saving RAM -- and allows even numbers of nodes... which brings joy :)