Make Unicode brackets opening/closing tokens (#18225)

This MR makes the Unicode bracket variants (⟦, ⟧, ⦇, and ⦈) act the same as their ASCII counterparts in the context of whitespace-sensitve symbols (see GHC Proposal #229 and MR !1664 (merged)). Because the implementation is character based instead of token based this must be done explicitly.

Note that the proposal does not actually include (| and |) in the definitions of opening and closing tokens, but the current implementation treats them as such so I've made sure their Unicode variants behave the same.