ALTO upload: Publish transcription on parents as well

Depends #148 (closed)

While we aggregate the text from the CONTENT attributes to build transcription, we should use the text attribute of a node's children when that doesn't work. In that case, confidence score of this aggregated transcription should be the micro-average of the children's (just like in #148 (closed)). We should resolve hyphenation when available, in SUBS_CONTENT attributes. These should be resolved only when porting transcription to the parent.

# arkindex_cli.commands.upload.alto.parser.AltoElement.text
from operator import attrgetter

    @property
    def text(self):
        """
        Easy access to the node's transcription
        """
        if not len(self.strings):
            return

        content = " ".join(string.attrib["CONTENT"] for string in self.strings).strip()

        if content:
            return content

        # Aggregate children `text` attribute
        # TODO: skip children with no transcription
        return "\n".join(map(attrgetter("text"), self.children))

Refs https://redmine.teklia.com/issues/6703