Hello.
A few comments on patch:
> + temporarily reducing the overhead of index maintenance
> + during bulk data loading operations
But tuples are still inserted, where the difference come from?
> or verifying an index is not being used
> + before dropping it
Hm, it does not provide the guarantee - index may also be used as an arbiter for INSERT ON CONFLICT, for example. For that case, "update pg_index set indisvalid = false" should be used before the DROP, probably.
Also index may also be used for constraint, part of partitioned table, etc.
Also, I think it is better to move check to indisvalid as if (!index->indisvalid || !index->indisenabled).
Best regards,
Mikhail.
