Arabic translation of ASVS v5.0.0

I would like to contribute an Arabic (ar) translation of ASVS 5.0.0.

I searched first: the only previous Arabic work is #1172 / #1174 against v4.0.3
in 2022, and there is no in-progress Arabic effort listed in CONTRIBUTING.md.

Status: complete. All 27 files in 5.0/en are translated, based on the
v5.0.0 branch as CONTRIBUTING.md requires, in a new 5.0/ar folder.

Checks:

• 5.0/tools/translation_validation/validate_translation.py --translated-folder ar
  reports Overall: PASS, 27/27 files, 0 requirement failures, 0 structural failures
• markdownlint clean against .markdownlint.jsonc
• All 345 requirement IDs and levels are byte-identical to the English
• The PR touches only 5.0/ar/ (28 files, insertions only)

Conventions used, following the merged translations rather than inventing new ones:

• Chapter/section headings keep the V<n> prefix and translate the name, as in tr/ru/id/ko/pt/fr
• Table headers translated (| # | الوصف | المستوى |)
• Glossary keeps the English term as the bolded lookup key with an Arabic gloss,
  following the Russian pattern - it keeps the A-Z ordering meaningful and stays
  searchable for practitioners who know the English terms
• Terminology inherited from the official 4.0.3 Arabic translation where it exists
  (37 of the 87 glossary terms), except for a handful of entries in 4.0/ar that
  are defective - PSTN and PII carry the wrong term, PBKDF2's definition is
  circular, and "Hibernate" was translated as "hibernation". Those were written fresh
• Protocol names and acronyms stay in Latin script (TLS, JWT, OAuth, SAML, WebRTC ...)
• Appendix C's algorithm tables are preserved verbatim - only headings, prose and
  table headers are translated. A transposed digit in an Argon2 parameter would be
  worse than leaving a row in English

One deliberate deviation from the English source: ≥ (U+2265) is written as >=
in Appendix C. Amiri has no glyph for U+2265, so it is silently dropped from the
PDF, turning m ≥ 47104 into m 47104 - a minimum reading as an exact value in a
password-hashing parameter table. See the separate issue on RTL PDF generation.

Happy to adjust any of the above - particularly terminology - if the working group
has preferences.

Use the v5.0.0 branch as required by CONTRIBUTING.md and translate the 27 files in 5.0/en into a new 5.0/ar folder. Run 5.0/tools/translation_validation/validate_translation.py --translated-folder ar and markdownlint with .markdownlint.jsonc; done means all 27 files pass structural and requirement checks, with IDs and levels unchanged.