[batch] Add deduped id column to resources table

[jigold]

This PR is the first step in getting a deduped resource id map into the database. The changes in this PR are to add a new column deduped_resource_id into the resources table and make sure that all new resources added to the database have the deduped_resource_id equal to the resource_id. We can do this because the mitigation that already merged #12654 and #12721 will assure that deduped_resource_id equals resource_id. The reason this PR needs to be here separately from the next one is because we need to have the batch service writing the new deduped_resource_id after the new rows have been inserted into the database. Unfortunately, we can't use a trigger for this because BEFORE INSERT triggers do not have the new value of an autoincrement column yet. We get the start of the newly inserted rows from the last insert id and then set all new ids from there equal to resource id that were NULL. We don't want to set all NULL rows to resource_id as we haven't take care of the old duplicate ids yet.

[daniel-goldstein]

I'm a little confused, could you update the PR body with a description of what the reasoning is here? How come the filter on resource_id > last_resource_id? Is deduped_resource_id = NULL not enough of a filter? And just to confirm, the point is to basically make a copy of the resource_id column into the deduped_resource_id column, but not for old rows. Why make a new column that is just a copy of another one?

I know this has been discussed a bit in zulip, but I think having the context in the PR body is very helpful for posterity and refreshing me on what's going on here.

[jigold]

Sorry about that! Let me know if the description isn't sufficient.

done

[daniel-goldstein]

Thanks for the explanation, makes more sense to me now. Looks good!