Check for BiDi markers in CLDR fields we split up

I have just noticed that the timezone L10n hourFormat field for Hebrew is bracketed with left-to-right BiDi markers. Since ldml.py splits the hourFormat (see LocaleScanner.timeZoneNames()), one half of it gets an initial marker and the other half gets the final marker. We need to check with folk who know glyph-rendering (e.g. esabraha) for what significance that might have – we may need to include both markers in both halves, for example – both for this case and any other Bidi marker characters we find in fields we munge in the course of extracting CLDR data. I know various Arabic locales, for example, have BiDi markers in some of the fields we extract, I just haven't previously thought about their relevance to how we extract and deploy the fields they mark.