Should "Juan Delos Reyes" read family 'Delos Reyes', as "Juan Dela Cruz" reads family 'Dela Cruz'? #1037

What happens. Delos is the one-word spelling of the Filipino particle de los (Delos Santos, Delos Reyes), the way Dela is of de la. fil.toml lists dela but not delos, so every configuration, the default included, reads it as a name word:

parse("Juan Dela Cruz")          →  given 'Juan', family 'Dela Cruz'
parse("Juan Delos Reyes")        →  given 'Juan', middle 'Delos', family 'Reyes'
parse("Ma. Luisa Delos Santos")  →  given 'Ma.', middle 'Luisa Delos', family 'Santos'

Bearers. In ORCID's own family-name box, Delos occurs 125 times among Philippine researchers (0.16% of 76,649) and 5 times in their given-name box. PatentsView has 5 Philippine family boxes. Delos is also a given name in the US (SSA 1,324+ births; 27 US citizens in Wikidata, e.g. Delos Wheeler). So it needs a given reading beside its particle reading, as dela has (dela keeps Dela, the short form of Adela).

The same family: las under a strict fil choice. es.toml lists las, but fil.toml does not, so parser_for(<fil>, strict=True) reads Maria Las Piñas as given 'Maria', middle 'Las', family 'Piñas'. It is 3 Philippine records in the strict-gap report (config_diff.py gaps fil, master f48c88a). #1031 added de, del, san, la and los and left las out.

Proposal. fil.toml gains delos (a particle written as one word, plus a given reading with its SSA count), and las, copied from es.toml. Measure with config_diff.py gaps fil and the snapshot.

Part of #753.