Should Catalan dels be a particle? Joan dels Reis reads middle dels, family Reis #1059

What happens today. dels (Catalan, "of the", plural) is in no vocabulary file, so it is read as an ordinary name word:

Maria dels Àngels Puig   →  given 'Maria', middle 'dels Àngels', family 'Puig'
Joan dels Reis           →  given 'Joan', middle 'dels', family 'Reis'

The first reading is right, but only by accident: it holds because dels is not a particle. Its singular del and the other particles chain forward into the family (rules/P2), so Maria de les Neus Puig reads family de les Neus Puig.

Why it was held back, and why that reason is gone. A #704 comment kept dels out of ca.toml, because making it a particle would move Maria dels Àngels Puig into the family, and the marked-word property moved nothing by default. Since #1058, rules/P10 (a given name a particle completes) reads a particle run plus a word marked after_particle as a middle name by default. Measured on master (2026-11-04) with Lexicon.default().add("dels", kind="article").add("àngels", kind="given", after_particle=True):

Maria dels Àngels Puig          →  middle 'dels Àngels', family 'Puig'        (unchanged)
Maria dels Àngels Puig i Vidal  →  middle 'dels Àngels', family 'Puig i Vidal' (unchanged)
Joan dels Reis                  →  family 'dels Reis'                          (moves)

Evidence so far (tools/name_usage.py --boxes, 2026-11-04):

• dels: ORCID 8 given boxes (Spain 5) and 2 family boxes, over the whole extract; none on the UK PSC register.
• àngels: ORCID 94 given boxes (Spain 51, Andorra 1) and 4 family boxes.

Both are thin, and the given-box dels records are the dels Àngels given name, not a surname.

To decide:

• Is there a population of Catalan surnames written with dels (dels Reis?) large enough to measure? Without one, the change protects nothing and moves no realistic name.
• Which kind dels would take (article, as del does in ca.toml).
• Whether àngels (and mercè, neus, carme) get after_particle. The ORCID bar decisions/O7 used (≥ 90%, ≥ 10 given records) is met by àngels alone, if at all.

Recorded as Open in decisions/P10. Split out of #704.