Should Smith Mary match Mary Smith? An opt-in match that reads one side under the other name order #1303

A list that mixes "First Last" and "Last First" without commas gives pairs the matcher calls inconsistent and keys apart. On master (01b982e):

>>> m = Parser().matching(other_readings=True)
>>> m.check("Smith Mary", "Mary Smith").consistent
False
>>> sorted(m.keys("Smith Mary")), sorted(m.keys("Mary Smith"))
(['mary;s'], ['smith;m'])

Callers work around it now by building a key for both orders: dataesr's athena builds prenom nom and nom prenom and confirms with an institution, and OpenAlex's pairwise check has a two-word swap fallback. ourresearch/whatsername#1 asks for the same thing on the parsing side.

Proposal: an opt-in "other reading" for the matcher: the name parsed again under the other name_order. It is tried the way other_readings tries an ambiguity's alternative, and keys() covers it, so keys still never separate a consistent pair. A match through it reads as its own outcome, so the table says one side was read in the other order.

Questions to settle

• Not on by default. Every name gets twice the keys, and Martin Thomas / Thomas Martin become consistent even when they are two different people.
• A second parse, not a field swap. Lim Guan Eng reads given/middle/family one way and family/given (Guan Eng) the other.
• Which orders count as "the other order." Under FAMILY_FIRST_GIVEN_LAST, a name has more than one alternative.
• Whether the parser should record an order fork itself. If it did, other_readings would reach this without a new switch. That ties in with the group order suggestion being designed for parsing (whatsername#2's list vote), Should an ALL-CAPS word in a mixed-case name mark the family name? DUPONT Jean gives given DUPONT #659 and Should Armitage JO read family 'Armitage'? A surname followed by initials, with no comma, reads the initials as the family name #1137.

Related: #1006, #1263.