0-mass fact

[BlakeMScurr]

I would be nice to have a fact that can group its children, always matches, and doesn't consume a node. Suppose we want to match sibling for loops somewhere in a function:

go.func_decl:
  go.nomass(depth = any):
    go.for_stmt
    go.for_stmt

If we used go.element instead of go.nomass we wouldn't match for statements that are direct children of the function.