`Josep Carod i Rovira` reads family `Rovira` — Catalan and Polish link surnames with `i`, which is not a conjunction

[derek73]

Catalan (and Polish) link two surnames with i, but i is not in CONJUNCTIONS, so the link is not joined:

"Josep Lluis Carod i Rovira"  ->  given='Josep'  middle='Lluis Carod i'  family='Rovira'
"Josep Carod i Rovira"        ->  given='Josep'  middle='Carod i'        family='Rovira'

The family should be the whole link (Carod i Rovira), which is how Spanish y already parses today:

"Juan Garcia y Lopez"         ->  family='Garcia y Lopez'

Why it was left out, and what measurement says

i is the Roman numeral I (suffixes.py:18), and the concern was the collision. Measured by adding i to a custom Lexicon's conjunctions, generation markers do not collide the way expected — every one of these is unchanged:

input	result	what protects it
John Smith I, john smith i	suffix='I'	a generation marker is trailing, and classified before grouping
John I Smith	middle='I'	P3's bare-Latin-capital veto: a capital single letter is an initial and never joins
john i smith	middle='i'	P3's three-word carve-out
Maier, Amy I, Jr.	unchanged	the only name in all three corpora (783 lines) with a bare i token

The collision is real, but it lands somewhere else

A word can hold both tags. With i added it carries conjunction and vocab:suffix, and _is_rootname() returns False for a suffix piece — so i shrinks the total that P3's carve-out tests (total < 4 in _group.py:200). It therefore needs one more name word than y does before it joins:

with i added to conjunctions:
"Josep Lluis Carod i Rovira"  ->  family='Carod i Rovira'   fixed
"Josep Carod i Rovira"        ->  family='Rovira'           NOT fixed
"Juan Garcia y Lopez"         ->  family='Garcia y Lopez'   (y counts itself toward total)

given + two surnames is the commonest Catalan shape, so adding the word alone fixes the longer names and leaves the common one behind.

What needs deciding

1. Add i at all — it serves Catalan and Polish.
2. If yes: should a word carrying both conjunction and vocab:suffix count toward the carve-out's rootname total? That is a C-ii question (set-vs-set collision, docs/design/decisions.md), and it is what decides whether Josep Carod i Rovira is fixed.
3. Out of reach either way: JOSEP LLUIS CAROD I ROVIRA keeps I an initial under the Latin-capital veto — that is Should Jose E Maria Santos join like Jose e Maria Santos does? (the single-letter connective's Latin-capital veto was never adjudicated) #383.

Docs affected

rules.md#P3's prose says the vocabulary holds six single-letter conjunctions (y e и і й و); this makes seven. decisions.md#3-0-reevaluations already records that which single letters a tradition wants joined is per-locale policy — i is another instance of it.