Use the symbo2verbalization mapping in lmh translate

[kohlhase]

So far lmh translate foo.tex en uses the translation mapping to translate e.g. \trefi[bar]{foo} to \mtrefi[bar?foo]{FOO} given that there is a pair <foo,FOO> In the translation mapping (\trefii and \trefiii are more complicated). This may lead to translations that are incorrect, in the face of polysemi and homonymy in verbalizations. What we should do instead is to use the symbol info in \trefi. So we should translate \trefi[bar]{foo} to \mtrefi[bar?foo]{FOO} given that there is a pair <bar?foo,FOO> in the symbol-to-verbalization mapping.

As the symbol-to-verbalization relation is ambiguous, we need to have some way of dealing with that. One way is always to just take the first one. A better one is to translate \trefi[bar]{foo} to \mtrefi[bar?foo]{FOO} OOOOOR \mtrefi[bar?foo]{FOOOOO} given that there are pairs <bar?foo,FOO> and <bar?foo,FOOOOO> in the symbol-to-verbalization mapping.

The symbol-to-verbalization mapping is generated as of MathHubInfo/Legacy-MathHub-Drupal#113