Improve metadata extraction handling

[tomdickman]

Currently, if metadata extraction fails, the same action is carried out every time (default adhoc task failure behaviour, increase faildelay and try again) but some resources may consistently fail extraction by a metadataextractor subplugin for some reason, either an edge case, or due to subplugin not correctly defining validation checks for a resource.

It would be nice if we could refactor the extraction process to require extractors to return an extraction status code too, along with their extraction response, so that more informative decisions can be made for failed extraction, based on whether this is a once off failure, or the resource is actually unprocessable by the extractor.

[tomdickman]

Important note, any changes should be backwards compatible, ie allow current metadataextractor subplugins to continue working without having to return a status code when extracting metadata while also supporting those which do.