tests: Make test_vm_bit_clear_on_heap_lock more robust again. (#6714)

When checking that the contents of the VM page in cache and in
pageserver match, ignore the LSN on the page. It could be different, if
the page was flushed from cache by a checkpoint, for example.

Here's one such failure from the CI that this hopefully fixes:
https://neon-github-public-dev.s3.amazonaws.com/reports/pr-6687/7847132649/index.html#suites/8545ca7650e609b2963d4035816a356b/5f9018db15ef4408/

In the passing, also remove some log.infos from the loop. I added them
while developing the tests, but now they're just noise.

test_runner/regress/test_vm_bits.py

@@ -167,10 +167,14 @@ def test_vm_bit_clear_on_heap_lock(neon_env_builder: NeonEnvBuilder):
 
     # The VM page in shared buffer cache, and the same page as reconstructed
     # by the pageserver, should be equal.
+    #
+    # Ignore the LSN on the page though (first 8 bytes). If the dirty
+    # VM page is flushed from the cache for some reason, it gets WAL-logged,
+    # which changes the LSN on the page.
     cur.execute("select get_raw_page( 'vmtest_lock', 'vm', 0 )")
-    vm_page_in_cache = (cur.fetchall()[0][0])[:100].hex()
+    vm_page_in_cache = (cur.fetchall()[0][0])[8:100].hex()
     cur.execute("select get_raw_page_at_lsn( 'vmtest_lock', 'vm', 0, pg_current_wal_insert_lsn() )")
-    vm_page_at_pageserver = (cur.fetchall()[0][0])[:100].hex()
+    vm_page_at_pageserver = (cur.fetchall()[0][0])[8:100].hex()
 
     assert vm_page_at_pageserver == vm_page_in_cache
 
@@ -201,16 +205,6 @@ def test_vm_bit_clear_on_heap_lock(neon_env_builder: NeonEnvBuilder):
     for _ in range(1000):
         cur.execute("select test_consume_xids(10000);")
     for _ in range(1000):
-        cur.execute(
-            "select get_raw_page_at_lsn( 'vmtest_lock', 'vm', 0, pg_current_wal_insert_lsn() )"
-        )
-        page = (cur.fetchall()[0][0])[:100].hex()
-        log.info(f"VM page contents: {page}")
-
-        cur.execute("select get_raw_page( 'vmtest_lock', 'vm', 0 )")
-        page = (cur.fetchall()[0][0])[:100].hex()
-        log.info(f"VM page contents in cache: {page}")
-
         cur.execute("select min(datfrozenxid::text::int) from pg_database")
         datfrozenxid = int(cur.fetchall()[0][0])
         log.info(f"datfrozenxid {datfrozenxid} locking_xid: {locking_xid}")