Clean up the replication tests (part of #3302) - #6714

- A slot with two-phase decoding enabled (PostgreSQL 15+) sends prepared
  transaction messages (b/P/K/r, and p when streaming) whatever the
  requested protocol version, but the message instances were only
  allocated for protocol version 3 and later, so protocol version 1
  and 2 sessions failed with a NullReferenceException. The same happened
  with protocol version 1 for stream aborts, which PostgreSQL 18.0 to at
  least 18.6 can send even without streaming and with any protocol
  version (bug #19616). All message instances are now allocated.
- Protocol version 4 sessions read every stream abort as if it carried
  the abort LSN and timestamp, which the server only sends with
  streaming = parallel. With streaming = on, and with streaming off when
  PostgreSQL 18 sent a stream abort anyway (bug #19616), this misparsed
  the stream. The layout is now chosen by the message length.
- GIDs of prepared transactions were read with synchronous I/O when they
  didn't fit into the buffered data, blocking a thread.
- Unknown message codes, unknown tuple types, unknown column kinds and
  changes for relations without a preceding Relation message now break
  the connection, like PostgreSQL's own apply worker errors out. Before,
  an Insert with an unknown tuple type was read as a new row, and in the
  other cases the exception left the connection in the Replication
  state, since nothing disposed the suspended replication enumerator:
  later commands on it threw NpgsqlOperationInProgressException, and the
  walsender kept the slot active until the connection was disposed. The
  errors for Delete and Truncate messages no longer say that they
  happened in an Update message. ReplicationConnection.DisposeAsync now
  also cleans up a replication enumerator that was left suspended when
  the connection broke.
- Reading a ReplicationValue after it had been consumed (e.g. the last
  column after the column enumeration ended) read whatever came next on
  the wire; it now throws InvalidOperationException, like
  TupleEnumerator.Current before the first MoveNextAsync. Reading a NULL
  as a nullable type no longer fails in text mode.
- The linked CancellationTokenSource created for each pgoutput and
  test_decoding enumeration is now disposed.
- PgOutputReplicationOptions.Equals and GetHashCode compared the list of
  publication names by reference; they now compare the names.

The pgoutput tests get a protocol version 4 fixture with streaming = on,
and the two-phase test now also runs with protocol versions 1 and 2.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

- The Stream, TextReader, ValueMetadata, Null and CompositeType pgoutput
  tests never disposed their replication connections, so every fixture
  leaked five walsenders until the test process exited, which eats into
  max_wal_senders.
- The pgoutput tests that used the same object names in all fixtures now
  use per-fixture names like the others, so that a test failing in one
  fixture can't affect the same test in the next one. Names too long for
  an identifier are now derived from a hash instead of a random GUID, so
  that the cleanup of the next run still finds what a killed run left
  behind.
- Tests that prepare a transaction name it like their slot, so that one
  a killed test left behind can be found and rolled back: by every
  logical fixture before its tests, and by the test's own cleanup.
  Creating a logical slot waits for every transaction in the cluster
  that holds an xid, so the next run hung at its first logical slot
  creation, long before the owning test, and the transaction's locks
  made the cleanup wait forever. TwoPhase is skipped when
  max_prepared_transactions is 0, like the distributed transaction
  tests.
- New TwoPhase_option test: no test sent pgoutput's two_phase option.
  It enables decoding prepared transactions, and the server enables
  two-phase decoding for the slot itself.
- New Stream_abort test: PostgreSQL 18 skips transactions that are
  already aborted when it would stream them, so the existing assertions
  never saw a stream abort there. The new test rolls back after part of
  the transaction was streamed.
- Replication_survives_pauses_longer_than_wal_sender_timeout no longer
  fails with negative intervals when wal_sender_timeout is disabled; it
  is skipped then, except on the build server.
- The devcontainer database gets the replication settings CI uses and a
  pg_hba entry for physical replication.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>