IssueTracker

Currently we do not have good benchmark / test coverage for clickable reuse across different configurations (e.g, passed interactionSource vs null interactionSource).

This might help identify if there are issues here, and allow for us to make more optimizations for the re-use case.

Some previous findings:

1. We don't undelegate the pointer input node when a node is reset / re-used, this might cause extra cost since we now have an extra node that might not be needed (but needs benchmarking)

2. When re-used, for correctness reasons we need to first remove the indication node in detach. When we are re-used, we first become attached, which causes us to create a new node and attach it, and _then_ we get the `update` call, which in most cases we need to remove the node we just attached, and create a new one. With the current node APIs I don't think there is a good way of resolving this issue as when we are attached we don't know if we are going to be updated or not, but this causes some unfortunate duplicate work for every re-use.