translation tooling reads and writes files using system Charset rather than UTF-8

JSON based translation repository files are read and written incorrectly. To reproduce:

• Add “-Dfile.encoding=US_ASCII” to your MAVEN_OPTS environment setting
• In for instance the Channel Manager, update a repository and Angular resource bundle and add a “é", build the Channel Manager
• In the hippo-cms-translations project, sub project Channel Manager, run “mvn hippo-cms-l10n:report” which should show the two changes, then register the changes using “mvn hippo-cms-l10n:update-registry”.
• Inspect the changed reference files, you’ll see that in both documents the “é” is serialized as “??"

I could not create a reproduction path for the reading bit, as that doesn't manifest itself independently afaics.

CSV import/export is done incorrectly. To reproduce:

• Add “-Dfile.encoding=US_ASCII” to your MAVEN_OPTS environment setting
• Create a UTF-8 import file, import it and you'll see that the changes are imported as "??"
• Do a full export of Chinese ("mvn hippo-cms-l10n:export-csv -Dlocale=zh -Dfull=true") and you'll see that all characters in the translations are ??