SAL's implementation of the I18nResolver.getAllTranslationsForPrefix() for JIRA is very inefficient

SAL's implementation of the getAllTranslationsForPrefix() method creates a new I18nBean every time which is very expensive. It should use the Caching Beanfactory to use a cached I18nBean instead. This means we need to introduce a new method on the I18nHelper interface to get all the keys starting with a particular prefix.