Making Programs Faster

`parse_text`

New tokenization:

     (This is a small stress test of the )
     (B<pod delimiter>)
     ( mechanism.  You\nare allowed to have )
     (X<< )
     (double >> and even )
     (Y<<< )
     (triple >>> delimiters.\nOrdinary )
     (Z<)
     (single )
     (I<delimiters>)
     ( may be> nested or may contain )
     (A<)
     (funny\n< characters>.  )
     (C<< )
     (Double )
     (D<delimiters>)
     ( may >> )
     (E<< )
     (also )
     (F<<< )
     (nest >>>\nif desired >>.\n)

So we now need to add handlers for the new X<complete sequence> tokens
In the old regime, the sequence would be put on the stack, then taken off again
- We'll just do that in one fell swoop