• fixes issues like http://builds-artifacts.piwik.org/piwik/plugin-Bandwidth/master/114/ where bandwidth is not available when report is flattened because Piwik applies queued filters to early and then does not know the mapping from metric id to metric name
• makes report fetching much faster. I don't remember exact numbers but I remember numbers like 20 to 30% that were spent in many cases because metrics needed to be renamed from metric id to metric name
• makes debugging easier as we can directly see the metric name instead of only metric ids
• makes developing faster because we remove a lot of complexity and when having new archiver we might not be able to use IDs anymore anyway because we cannot really give IDs dynamically

Should not use too much storage because we gzip the reports anyway when storing data tables. The only "disadvantage" I can think of is that we cannot rename metrics as easily anymore (it would be currently only a change in the mapping of metric id to metric name), on the other side we could still rename metrics when applying queued filters so it shouldn't be a big issue.