Faster forms of truncate are not yet implemented for tiered storage. At a shallow level, this is because tiered tables are "table:", and underneath that, "tiered:", and so are not recognized as "file:". Non-tiered tables have "file:" underneath them. For purposes of truncate, "file:" uris are recognized specially and allow __wt_btcur_range_truncate to be called. With tiered tables, we fall back to a default using generic cursor operations, basically walking the cursor calling WT_CURSOR->remove for each item.

This could probably be fixed fairly easily, then we would have fast truncate for the first stage of tiered storage (without sharing). However, for tiered storage with sharing, we would use union tables, and truncating a range with union tables would require more thought (maybe using a tombstone that has a range?).

We should first evaluate whether we need a "tiered:" fast delete, or a "union" fast delete or both.