Speaker identification as default templated curly bracket parsing

[lmmx]

• Default interpretation of curly brackets set by template, but for most current purposes it will be speaker identification
  • -{abbrev ⠶ full name, role} or -{full name ⠶ abbrev} (should probably standardise this, oops)
    • [usually alongside first speech]
  • -Name {abbrev} ⇢ -..metadata attachment format
    • usually multiple listed together, prefacing speech as 'dramatis personae'

[lmmx]

Currently not parsed, e.g. see Q&A lines

-{BJ}
-:Thanks, Beth
-Look, I just really repeat the point I made to Laura.

This should use a Q&A template and the curly bracketed line will be some sort of block-level prefix, making the line "Thanks, Beth" become the 'first line' of the block, therefore triggering a search for the preceding question to tie the answer to

• Note that this clashes with the interpretation of the config setting
  ALLOW_LIST_WITHOUT_SUFFIX_INIT (True) in tokens.py.
• This clash should be worked out somehow, as usually it's only fair to tag questions to answers at the block level (and this is consecutive blocks)
  • Perhaps it'd be sufficient to just rely on a question mark suffix on the line (or a 'conjoined' line), but I have not yet implemented logic to indicate 'conjoined' nodes