Show more of the parser in the README example - #340

The 1.4 example showed the whole parse at once through the repr; 2.x
had shrunk to two attribute accesses. The repr is back, since it is the
only thing that shows every field in one place, and family_base /
family_particles come with it because the particle split is a 2.x view
1.4 had no equivalent for.

Six scripts now, all under the default configuration: Chinese and
Japanese family-first, the Korean census split, a glued honorific,
a Cyrillic title, and the Arabic bn chain. Russian patronymics go last
because they are the one case here that needs its pack, which is worth
showing rather than hiding.

No subheadings and no explanatory prose. The comments carry it.

Dropped the paragraph naming the seven fields: the intro two
paragraphs up already names them, and the repr now shows them.

Every line is doctested by the README doctest CI step. Verified the
file is still valid RST and that twine accepts it as a long
description, since PyPI renders it.

The example was '毛 泽东', spaced, which is not how Chinese names are
written. It was chosen because the spaced form demonstrates
family-first order without needing a pack, but that is a capability
shown against input nobody has: a reader who tries the realistic
'毛泽东' gets the whole name in family and concludes it does not work.

It now uses '毛泽东' under locales.ZH, which is both the real spelling
and the real answer, and it sits beside the Russian patronymic because
they need a pack for the same class of reason.

That grouping is the honest one. Korean unspaced works by default
because hangul is written by one language and its surnames are a
closed census set; Han text cannot say which language wrote it, so a
Chinese surname list would split 高橋一郎 in the wrong place. The
comment says so.