Leverage same mechanism used by surrounding pairs in token whitespace removal logic

[pokey]

The problem

As of #1564, we determine whether we should remove whitespace in part based on whether the remaining adjacent token is an opening / closing pair. This mostly works, but has a couple problems:

• We duplicate our source of truth about which tokens are considered opening / closing pairs
• The logic breaks down in complex cases, eg when it's not obvious whether a " is opening or closing (see test case below)

The solution

We should find a way to leverage the code we use in surrounding pairs to determine if a token is opening or closing

Dependencies

• Migrate surroundingPair to use a scope handler #1061

Test cases to add

The following test case fails today, but should pass:

languageId: plaintext
command:
  version: 5
  spokenForm: chuck token
  action: {name: remove}
  targets:
    - type: primitive
      modifiers:
        - type: containingScope
          scopeType: {type: token}
  usePrePhraseSnapshot: true
initialState:
  documentContents: "\"aaa\".length + 1"
  selections:
    - anchor: {line: 0, character: 5}
      active: {line: 0, character: 12}
  marks: {}
finalState:
  documentContents: "\"aaa\" + 1"
  selections:
    - anchor: {line: 0, character: 5}
      active: {line: 0, character: 5}