In disagg, ingest tables don't use tombstones, but use a special wt_tombstone value instead.
This causes a problem because during step up, we perform prepare discover process and can claim a prepared tombstone value, causes a tombstone to be accidentally committed on the ingest table.

This can cause issue when look up:

• when going down the update chain and encountering a tombstone on the ingest cursor, ingest cursor will return a WT_NOTFOUND error, meaning the key is no longer visible.
• However WT_NOTFOUND is also returned when ingest table cannot find any update at all, which we would then start searching for stable table -> it can return a deleted value. Therefore we cannot use a tombstone type on ingest table.

This PR fixes the issue by replacing the tombstone to an update with special tombstone value instead of a real tombstone update structure when finding pending prepared items.