Nickel: Start infix-chains also for annotated infix_expr

[ErinvanderVeen]

This resolves part of #547, in particular the issue investigated by @torhovland.

[
  a && (b & c | d),
  a &&
  (b & c | d),
  a &&
  (b &
  c | d)
]

Get correctly formatted as:

[
  a && (b & c | d),
  a
  && (b & c | d),
  a
  && (
    b
    & c | d
  )
]

@torhovland's reasoning was correct, there was a scoping issue. The scope was started by the [, ending after the infix expression. Since the [ was on a previous line, the entire expression was considered to be multi-line, and thus expanded as such. The scope was started there (and not on the next line) because that line was considered a annotated_infix_expr instead of an infix_expr which we did not include in our query to start the scope.

[ErinvanderVeen]

Also, this PR makes it so that the debug messages for appending and prepending are correct. Previously these messages always reported Empty for the Atom, since the Atom was removed. For our sanity it also removes the message entirely when it is empty.