Is there an existing issue for this?

• I have searched the existing issues

New feature

We know that the applicator is getting in to a blocked state and is unable to make progress. We have identified several issues causing this behavior, but are still receiving reports. This is complicated code and will have a long tail to fix every bug.

In the meantime, we should add a 'reset' function on the applicator that is called when gossip is toggled off. This will fix the applicator and allow the node to sync and resume normal function.

When reset is called, the applicator should also dump its internal state so that we can inspect what caused the problem, which will aid in future debugging.

Anything else?

No response