Latin semicolon (003B) with Urdu text incorrectly renders as an Arabic semicolon (061B)

[saadatm]

If U+003B SEMICOLON is used with some Urdu text, its shape is rendered as U+061B ARABIC SEMICOLON:

hb-view -o ur_en-semicols.png Gulzar-Regular.ttf "ایک; دو; تین"

This is problematic because the use of the Latin semicolon (003B) in Urdu is a mistake; authors should use the Arabic semicolon (061B) instead. Gulzar using the Arabic shape for the Latin semicolon makes this mistake harder to spot. Also, if an author does intend to use the Latin semicolon in Urdu text (for… I don’t know, some instructional text, maybe?), then the rendering still comes out wrong.