Refine Locale Display Name Algorithm to handle names for aliased subtags
The spec for Locale Display Name Algorithm needs to be refined to properly handle distinct names for aliased subtags, as in the case of "nb" vs "no" (it needed some clarification anyway).
The following statement
First, canonicalize the locale identifier as per Part 1, Section 3.2.1 Canonical Unicode Locale Identifiers.
needs to be clarified as follows:
First, convert the locale identifier to canonical syntax per ...
2. In the subsection Processing types of locale identifier subtags, the step for Language currently begins:
Match the L subtags against the type values in the <language> elements. Pick the element with the most subtags matching. If there is more than one such element, pick the one that has subtypes matching earlier. If there are two such elements, pick the one that is alphabetically less.
We need to add another sentence after the above:
If there is no match, then further convert L to canonical form per Part 1, Section 3.2.1 Canonical Unicode Locale Identifiers and try again.
ICU implementation to be in