pgsql: Allow relkind 'I' in type_sanity test.

Allow relkind 'I' in type_sanity test.

This is cosmetic at present, since no partitioned indexes have
been created yet at the stage of the regression tests when
type_sanity runs. But it's confusing that this list of allowed
values doesn't match reality, so fix it.

Noted by Jian He.

Discussion: /messages/by-id/CACJufxHhr58q-YoBeqnY5P-u8Xe2X5sMtsr3LYE1nQDwdqjB4w@mail.gmail.com

Branch
------
master

Details
-------
https://git.postgresql.org/pg/commitdiff/4b9fbd6be442cf36910fb34308e0edfd43f2dbdf

Modified Files
--------------
src/test/regress/expected/type_sanity.out | 2 +-
src/test/regress/sql/type_sanity.sql | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)