Should de written in Sinhala or Tamil script join the family name, as Latin de does? (අජිත් ද සිල්වා gives middle ද, family සිල්වා) #1245

Sri Lankan surnames of Portuguese origin keep their de in every script: Ajit de Silva is අජිත් ද සිල්වා in Sinhala and அஜித் டி சில்வா in Tamil. si.toml lists the Latin de, so the Latin name keeps it in the family, but neither native spelling is listed, so P2 (particles join forward) doesn't reach them (measured on master ccaf6ad):

How often (Wikidata labels of Sri Lankan citizens, 1,185 Sinhala and 1,520 Tamil, 2026-11-08): Sinhala ද stands before the last word in 101 labels, every one of them de/De/da in the English label (de Silva, de Zoysa, de Mel, de Alwis); all 101 read middle ද. Tamil டி does so in 50, and all 50 read middle டி. The Latin de reads family in 120 of 120 of the same people's English labels.

The Tamil catch: டி is also how Tamil spells the letter D, so டி. is the initial "D." (26 labels, 17 of them opening the name). The fold strips the period, so a listed டி would be read in டி. too. In a lexicon of my own with டி as a particle, டி. எஸ். டி சில்வா reads family டி. எஸ். டி சில்வா whole (P1, the lone particle fold). Sinhala has no such clash: its D is ඩී, and ද. appears as an initial in only one label.

Questions

• Should Sinhala ද join si.toml's de reading (a preposition, never a given name)?
• Should Tamil டி be listed too, given that its dotted form is an initial? A particle reading that stops at the period would need the fold to tell டி from டி.; does anything else need that?
• Other spellings I saw once or twice: Tamil தி (ஆர். எல். தி கரிஸ்டர்), Sinhala ද. written with a period. Leave those out?