Summary:
This change updates test files within `TestPgRegressPartitions` where plan outputs were changed due to the following reasons:
- Introduction of Incremental Sort
- For a partitioned table, Upstream PG seems to create a separate relid for the parent of all the partitions as of commit 55a1954da16e041f895e5c3a6abff13c5e3a4a2f. This causes explain output changes when partitioned tables are involved.
- Upstream commit 86dc90056dfdbd9d1b891718d2e5614e3e432f35 made it such that the an UPDATE/DELETE/INSERT node estimates 0 rows for itself in the absence of a RETURNING clause.
- Remove Append node in some EXPLAIN plans: upstream PG 8edd0e79460b414b1d971895312e549e95e12e4f. This caused some cost changes too.
- Upstream commit 9fdb675fc5d2de825414e05939727de8b120ae81 changes how partition pruning is done. This caused some differences in partitions that were pruned, even "unpruning" some partitions in a few cases. Cases where partitions were unpruned were verified to behave similarly to upstream postgres.
- Upstream commit c5b7ba4e67aeb5d6f824b74f94114d99ed6e42b7 fixes error messages during updates into partitioned rels to be more consistent with the given query. For example, there is a change in yb_pg_update.out:154 that fixes the error message to refer to the relation given by the query and the failing row format also now exactly matches the actual table's column format.
Jira: DB-13187, DB-9525

Test Plan:
Jenkins: test regex: .*Partition.*
./yb_build.sh --java-test 'org.yb.pgsql.TestPgRegressPartitions'

Reviewers: jason, telgersma

Reviewed By: jason

Subscribers: smishra, kramanathan, yql

Differential Revision: https://phorge.dev.yugabyte.com/D37597