Handle update of VM in XLOG_HEAP_LOCK/XLOG_HEAP2_LOCK_UPDATED WAL records (#4896)

## Problem

VM should be updated if XLH_LOCK_ALL_FROZEN_CLEARED flags is set in
XLOG_HEAP_LOCK,XLOG_HEAP_2_LOCK_UPDATED WAL records

## Summary of changes

Add handling of this records in walingest.rs

## Checklist before requesting a review

- [ ] I have performed a self-review of my code.
- [ ] If it is a core feature, I have added thorough tests.
- [ ] Do we need to implement analytics? if so did you add the relevant
metrics to the dashboard?
- [ ] If this PR requires public announcement, mark it with
/release-notes label and add several sentences in this section.

## Checklist before merging

- [ ] Do not forget to reformat commit message to not include the above
checklist

---------

Co-authored-by: Konstantin Knizhnik <knizhnik@neon.tech>

test_runner/regress/test_vm_bits.py

@@ -111,3 +111,103 @@ def test_vm_bit_clear(neon_simple_env: NeonEnv):
     assert cur_new.fetchall() == []
     cur_new.execute("SELECT id FROM vmtest_cold_update2 WHERE id = 1")
     assert cur_new.fetchall() == []
+
+
+#
+# Test that the ALL_FROZEN VM bit is cleared correctly at a HEAP_LOCK
+# record.
+#
+def test_vm_bit_clear_on_heap_lock(neon_simple_env: NeonEnv):
+    env = neon_simple_env
+
+    env.neon_cli.create_branch("test_vm_bit_clear_on_heap_lock", "empty")
+    endpoint = env.endpoints.create_start(
+        "test_vm_bit_clear_on_heap_lock",
+        config_lines=[
+            "log_autovacuum_min_duration = 0",
+            # Perform anti-wraparound vacuuming aggressively
+            "autovacuum_naptime='1 s'",
+            "autovacuum_freeze_max_age = 1000000",
+        ],
+    )
+
+    pg_conn = endpoint.connect()
+    cur = pg_conn.cursor()
+
+    # Install extension containing function needed for test
+    cur.execute("CREATE EXTENSION neon_test_utils")
+
+    cur.execute("SELECT pg_switch_wal()")
+
+    # Create a test table and freeze it to set the all-frozen VM bit on all pages.
+    cur.execute("CREATE TABLE vmtest_lock (id integer PRIMARY KEY)")
+    cur.execute("INSERT INTO vmtest_lock SELECT g FROM generate_series(1, 50000) g")
+    cur.execute("VACUUM FREEZE vmtest_lock")
+
+    # Lock a row. This clears the all-frozen VM bit for that page.
+    cur.execute("SELECT * FROM vmtest_lock WHERE id = 40000 FOR UPDATE")
+
+    # Remember the XID. We will use it later to verify that we have consumed a lot of
+    # XIDs after this.
+    cur.execute("select pg_current_xact_id()")
+    locking_xid = cur.fetchall()[0][0]
+
+    # Stop and restart postgres, to clear the buffer cache.
+    #
+    # NOTE: clear_buffer_cache() will not do, because it evicts the dirty pages
+    # in a "clean" way. Our neon extension will write a full-page image of the VM
+    # page, and we want to avoid that.
+    endpoint.stop()
+    endpoint.start()
+    pg_conn = endpoint.connect()
+    cur = pg_conn.cursor()
+
+    cur.execute("select xmin, xmax, * from vmtest_lock where id = 40000 ")
+    tup = cur.fetchall()
+    xmax_before = tup[0][1]
+
+    # Consume a lot of XIDs, so that anti-wraparound autovacuum kicks
+    # in and the clog gets truncated. We set autovacuum_freeze_max_age to a very
+    # low value, so it doesn't take all that many XIDs for autovacuum to kick in.
+    for i in range(1000):
+        cur.execute(
+            """
+        CREATE TEMP TABLE othertable (i int) ON COMMIT DROP;
+        do $$
+        begin
+          for i in 1..100000 loop
+            -- Use a begin-exception block to generate a new subtransaction on each iteration
+            begin
+              insert into othertable values (i);
+            exception when others then
+              raise 'not expected %', sqlerrm;
+            end;
+          end loop;
+        end;
+        $$;
+        """
+        )
+        cur.execute("select xmin, xmax, * from vmtest_lock where id = 40000 ")
+        tup = cur.fetchall()
+        log.info(f"tuple = {tup}")
+        xmax = tup[0][1]
+        assert xmax == xmax_before
+
+        if i % 50 == 0:
+            cur.execute("select datfrozenxid from pg_database where datname='postgres'")
+            datfrozenxid = cur.fetchall()[0][0]
+            if datfrozenxid > locking_xid:
+                break
+
+    cur.execute("select pg_current_xact_id()")
+    curr_xid = cur.fetchall()[0][0]
+    assert int(curr_xid) - int(locking_xid) >= 100000
+
+    # Now, if the VM all-frozen bit was not correctly cleared on
+    # replay, we will try to fetch the status of the XID that was
+    # already truncated away.
+    #
+    # ERROR: could not access status of transaction 1027
+    cur.execute("select xmin, xmax, * from vmtest_lock where id = 40000 for update")
+    tup = cur.fetchall()
+    log.info(f"tuple = {tup}")