[Lexicon/Wiki] Handle Contextual Arabic Letter Forms

[remnqymi]

Our transcription of Crum uses Arabic Presentation Forms-B, which can cause search to fail because queries are expressed in normal letters. For example, compare the queries خ and ﺧ.

The presentation forms exist in the Unicode only for compatibility with older standards, and are not currently needed for coding text. They're essentially deprecated, however, we have chosen to use them in our transcription!

IDEA:

• The pipeline should ensure that Arabic presentation forms-B is not used except where strictly needed and explicitly noted in encoding decisions.
• Perform some form of normalization in order for search to succeed regardless of the encoding.

IDEA:

• Abandon Arabic Presentation Forms-B. Mimic the appearance in Crum's book by using Arabic Tatweel.

IDEA:

• Abandon Arabic Presentation Forms-B. Mimic the appearance in Crum's book by changing the font to render Arabic letters in contextual form when followed by Arabic Fatha Isolated Form, which we use to mark abbreviations. (This may be the cleanest fix. We shouldn't continue to use deprecated Unicode points, neither should we add filler characters to the encoding.)