BUG #17959: amcheck fails to find a matching index tuple for an invisible heap tuple

The following bug has been logged on the website:

Bug reference: 17959
Logged by: Alexander Lakhin
Email address: exclusion(at)gmail(dot)com
PostgreSQL version: 16beta1
Operating system: Ubuntu 22.04
Description:

After some DDL/transactional operations (a reproducer to follow)
pg_amcheck detects an anomaly:
btree index "regress001.pg_catalog.pg_depend_reference_index":
ERROR: heap tuple (13,35) from table "pg_depend" lacks matching index
tuple within index "pg_depend_reference_index"

The corresponding table heap page contains:
lp | lp_off | lp_flags | lp_len | t_xmin | t_xmax | t_field3 | t_ctid |
t_infomask2 | t_infomask | t_hoff | t_bits | t_oid |
t_attrs |
raw_flags |
combined_flags
35 | 6232 | 1 | 49 | 734 | 736 | 2 | (13,35) |
8199 | 1280 | 24 | | |
{"\\xeb040000","\\x38400000","\\x00000000","\\x370a0000","\\x36400000","\\x00000000","\\x6e"}
| {HEAP_XMIN_COMMITTED,HEAP_XMAX_COMMITTED,HEAP_KEYS_UPDATED}
| {}

pg_depend_reference_index contains:
itemoffset | ctid | itemlen | nulls | vars | data
| dead | htid |
157 | (13,35) | 24 | f | f | 37 0a 00 00 36 40 00 00
00 00 00 00 00 00 00 00 | t | (13,35) |

SELECT ctid, * FROM pg_depend WHERE refclassid = 0x0a37 AND refobjid =
0x4036 AND refobjsubid = 0
doesn't return any rows.

Shouldn't amcheck ignore invisible tuples?