I reviewed completely how pronunciations were retrieved in order
to tackle that task, including changing where pronunciations were
looked for: before we were using the whole word wikicode, now we
only use `head_sections` wikicode (faster, and more efficient).

It has the following nice side-effects (on top the having
local-specific formatting like `\...\` for French, `[...]` for German, etc.):

- `EN`, `FR`: now supports multiple cross-sections pronunciations
  (like ones from Middle English combined to ones from Old English
  sections, for instance on English data)
- `FR`: pronunciations outside the French section are no more taken
  into account (like using values from Gaulois or other locale sections,
  that was incorrect)
- `PT`: a lot more pronunciations are correctly found now
- `SV`: multiple pronunciations are now supported
- `RU`: it will be way easier to tackle #1376 then

I reviewed completely how pronunciations were retrieved in order
to tackle that task, including changing where pronunciations were
looked for: before we were using the whole word wikicode, now we
only use `head_sections` wikicode (faster, and more efficient).

It has the following nice side-effects (on top the having
local-specific formatting like `\...\` for French, `[...]` for German, etc.):

- `EN`, `FR`: now supports multiple cross-sections pronunciations
  (like ones from Middle English combined to ones from Old English
  sections, for instance on English data)
- `FR`: pronunciations outside the French section are no more taken
  into account (like using values from Gaulois or other locale sections,
  that was incorrect)
- `PT`: a lot more pronunciations are correctly found now
- `SV`: multiple pronunciations are now supported
- `RU`: it will be way easier to tackle #1376 then