Thanks @deherman-MSFT .
The confusion is because the documentation states that the metadata will be left on line for discovery, when in fact the search is not able to use that metadata even if 'use storage metadata only' option is ticked.
From what I can tell I can index the blob files whilst the files are in cool or hot tiers, then keep the index active to allow search to identify them in the index even when they have moved to archive tier. We will be using Azure lifecycle management to move from cool to archive, so all blobs will be indexable initially. If you can comment on if this approach sounds sensible that would be appreciated, thanks
Ivan