Get a dictionary for the specified writing system, or null if we don't know of one. We ideally want a dictionary that exactly matches the specified writing system, that is, the file name for the .dic file == the SpellCheckDictionary of the writing system. If we can't find such a dictionary, for major languages (those we didn't create from wordform inventory), we will return a dictionary that shares a prefix, for example, 'en' when looking for 'en_US' or vice versa. This is not allowed for vernacular languages (where the dictionary is one we created ourselves); we return null if we can't find an exact match or an approximate match that is a 'major' language dictionary.