CLDR-15611 Update tr35.md for special script codes (#4061)

docs/ldml/tr35.md

@@ -728,25 +728,38 @@ the Visual variants are distinct in appearance, but otherwise encompass a single
 and the Subsets exclude certain characters from a script.
 The Equivalents for Subsets are not as well defined, so the "Equivalents" are marked as approximate.
 
-| Group | Script | Equivalent |
+| Variant | Script | Equivalent |
 | --- | --- | --- |
-| Compounds | Jpan | ≡ Hani ∪ Hira ∪ Kana |
+| Compound | Jpan | ≡ Hani ∪ Hira ∪ Kana |
 |  | Hrkt | ≡ Hira ∪ Kana |
 |  | Kore | ≡ Hani ∪ Hang |
 |  | Hanb | ≡ Hani ∪ Bopo |
-| Visual variants | Aran | ≡ Arab (Nastaliq variant) |
+| Visual | Aran | ≡ Arab (Nastaliq variant) |
 |  | Cyrs | ≡ Cyrl (Old Church Slavonic variant) |
 |  | Latf | ≡ Latn (Fraktur variant) |
 |  | Latg | ≡ Latn (Gaelic variant) |
 |  | Syrn | ≡ Syrc (Eastern variant) |
 |  | Syre | ≡ Syrc (Estrangelo variant) |
 |  | Syrj | ≡ Syrc (Western variant) |
-| Subsets (approximate) | Jamo | ≃ Hang - LVT - LV |
-|  | Hans | ≃ Hani - Traditional-only |
-|  | Hant | ≃ Hani - Simplified-only |
+| Subset | Jamo | ≃ Hang − LVT - LV |
+|  | Hans | ≃ Hani − Traditional-only |
+|  | Hant | ≃ Hani − Simplified-only |
 
-The special codes most frequently used are in the locale identifiers zh-Hans, zh-Hant, ja-Jpan, and ko-Kore.
+The special codes most frequently used are in the locale identifiers `zh-Hans`, `zh-Hant`, `ja-Jpan`, and `ko-Kore`:
+the first two are **Subsets**, and the last two are **Compounds**.
 These are used, for example, in [Likely Subtags](#Likely_Subtags) in LDML.
+
+The **Equivalent** values in the **Subset** variants are only approximate, _and_ the variants are also visual variants.
+Thus `Hans` is a request for:
+- Not using characters that are Traditional-only
+- Characters common between Simplified and Traditional to be given a Simplified rendering.
+
+**Visual** variant script codes (that are not **Subset** variants) can be used in a locale identifier to request a particular rendering.
+For example, ar_Aran could be used to request that ar_Arab data be used, but with a Nastaliq-style font.
+However, the few variant script codes represent only a very small fraction of the different script variants in use.
+Moreover, this feature is not widely supported, and may give unexpected results when not supported.
+For example, an implmentation might not recognize `Aran` in `uz-Aran` at all, and return results for `uz-Latn`. 
+
 Some of the special codes are used in other specifications,
 such as in [Mixed_Script_Detection](https://unicode.org/reports/tr39/#Mixed_Script_Detection).