Don't use ctid scan for negative relpages (like during scanning a partitioned table).

[simi]

Scanning a partitioned table caused an infinite hang because PostgreSQL sets relpages=-1 for relkind='p' tables. Cast to idx_t (uint64) gives pages_approx = 2^64-1, so the scanner loops over ~18 quintillion pages.

Now throws NotImplementedException at bind time with a message that includes the relation kind and table name. Now fallback to 0 estimate as scanner already does in few other cases. Keep original relpages value (idx_t needs to be "downgraded" to int64_t to be able to hold original -1 value) in approx_num_pages and later during preparing scan, check for negative approx_num_pages and fallback to use_ctid_scan = false for now.

---

Initial step to provide support for partitioned tables, since those are broken as reported at #220.

I have plan to open set of smaller PRs to improve the situation ending with full support of partitioned tables. This is initial foundation - accepting the broken state and fallback to slower but working scan.

Example EXPLAIN (SELECT hangs):

lake D EXPLAIN (FORMAT JSON) SELECT * FROM pg.public.sales_revenues;

┌─────────────────────────────┐
│┌───────────────────────────┐│
││       Physical Plan       ││
│└───────────────────────────┘│
└─────────────────────────────┘
[
    {
        "name": "POSTGRES_SCAN",
        "children": [],
        "extra_info": {
            "Table": "sales_revenues",
            "Projections": [
                "id",
                "account_id",
                "sale_id",
                "report_id",
                "report_group_id",
                "organization_id",
                "contract_id",
                "configuration_id",
                "product_id",
                "product_asset_id",
                "asset_id",
                "base",
                "share",
                "royalty",
                "currency",
                "chain",
                "statement_id",
                "unit_amount"
            ],
            "Estimated Cardinality": "18446744073709551568"
        }
    }

pg_class returns

reporting_development=# SELECT relkind, relpages FROM pg_class WHERE relname = 'sales_revenues';
 relkind | relpages 
---------+----------
 p       |       -1

[simi]

I'll check on CI failures.

[staticlibs]

@simi

Just FYI, some details on local test runs: #431 (comment)

[simi]

@staticlibs thanks for the hint, by mistake I have run only new tests before submitting PR. Clearly this should not break ATTACH. I have reworked this batch to just fallback to use_ctid_scan = false for negative relpages (as partitioned table provides). I have added new commit to see the difference over original idea and updated the PR name and description. I can squash if needed. If welcomed, I'll continue working on better support of partitioned tables in other PRs.

[staticlibs]

Thanks! Looks good to me.