Today, if you clone a sensor (e.g. from the management view of the device tree), all channels from the cloned sensor get copied over too. This can result in erroneous channels on the new copy.
For example, if I create a CPU load sensor on server A that has 8 procs, then clone that sensor onto server B that only has 4 procs, the B sensor will still show 8 channels. This is a trivial example and possibly easy to ignore these extra 4 CPU channels, but this becomes much much worse for other types of sensors and as the channel count increases.
In my opinion, any time a sensor is cloned to a new device all channels should be cleared so that we can gather clean data. Otherwise, this makes cloning sensors almost completely useless for me. I have to re-create every sensor on new devices manually so that I can be sure I get the right channels. If channels were cleared on cloning, it would drastically decrease set up time for new devices/sensors.