Fetching large ACS's fail unless a large timeout is specified

[da-tanabe]

We've observed that when fetching unusually large ACS'es, dazl will terminate the connection with a failure if it takes more time than the default timeout (30 seconds) to pull down.

There's a chance that this code isn't doing what I think it should be doing:

dazl-client/python/dazl/ledger/grpc/conn_aio.py

Lines 1237 to 1246 in 432003b

	# Unidirectional gRPC streams cannot sensibly have a deadline because the stream may be
	# open indefinitely. However, if fetching an individual message from the stream takes a
	# long time here, we can reasonably assume that the stream is dead, because Active
	# Contract Set messages are really supposed to be sent as quickly as the server can send
	# them. In other words, a long timeout pause in the middle of pulling down ACS messages
	# would be highly unusual, so we treat them as fatal
	i = response_stream.__aiter__()
	while True:
	try:
	response = await wait_for(i.__anext__(), timeout=self._call.timeout_seconds)

Determine why larger timeouts seem to allow this code to run successfully.