Splitting steps corrupts step_count

[rusty-snake]

This is a follow-up on #8069, so ping @matkoniecz @quincylvania @DerDings @westnordost

---

The current behaviour of iD when splitting ways is to create two ways with identical tags. So if you have a 10m footway with surface=asphalt but actually the first half is surface=pacing_stone, you place a node, split the way and change the surface. The same applies for all other kinds like access*, smoothness, cycleway, tracktype, ... Except for step_count which gets very silently modified. If you split a highway=steps to change a steps related tag such as width, this is good (i.e. no double count). If you however go to change one of the two ways to a highway=footway because there was just a long highway=steps mapped where half of the distance is actually only the footway to it, you end up with

1. a highway=footway with step_count=*. If it has steps, it shouldn't be a footway but if it does not have steps, the tag does not make sense.
2. a highway=steps with an incorrect step count. Sometimes even step_count=0.

So especially in the case of changing a highway=steps to a highway=footway|path it would be good to nuke the step_count, mark it for re-survey or somehow notify the user that they did not got an identical copy of the tags.

[matkoniecz]

The current behaviour of iD when splitting ways is to create two ways with identical tags.

per #8069 (comment) and d43fc62 step_count is handled differently

has it changed

[rusty-snake]

Comes later:

Except for step_count which gets very silently modified.

It didn't change. The problem is that is happens very silently and can hurt sometimes. As it the only(?) tag that gets modified by the split action it happens unexpected.