fix-7560Add Vietnamese collation benchmark

[Jayant-kernel]

Adds Vietnamese name-like strings benchmark for collation performance testing.
The benchmark uses common Vietnamese surnames (Nguyễn, Trần, Lê, Phạm, etc.)
with various diacritical marks to provide interesting Latin1/UTF-16 performance
comparisons.

Fixes #7560

[CLAassistant]

All committers have signed the CLA.

[Jayant-kernel]

@hsivonen @echeran
review this correct one

[hsivonen]

Thanks. The data superficially (I don't know enough Vietnamese to say for sure) appears to follow Vietnamese name structure, but it's not actually useful for exercising Vietnamese collation, because the first difference is never the same base vowel with different diacritics. Also, the list seems to over-do the identical prefixes.

[Jayant-kernel]

@hsivonen @sffc @echeran
Thanks for the feedback! I've completely revised the benchmark data.

What Changed:
Focus on diacritical mark differences on the same base vowel
Complete coverage of all 5 Vietnamese tone marks (sắc, huyền, hỏi, ngã, nặng)
Pattern examples: ba → bá → bà → bả → bã → bạ | ban → bàn → bán → bản → bạnIncludes ~30+ actual Vietnamese syllables (bàn=table, bạn=friend, ăn=eat, etc.)
12 consonant varieties to prevent prefix over-repetition

Result:
Now properly tests Vietnamese collation where the first difference is always a tone mark on the same base character, exercising Latin1/UTF-16 performance with diacritics.