Hi, I've been trying to use the Python Pulsar client with a bare metal cluster and I've been finding that has_message_available() is much slower from 3.5.0 onwards.

For example, if I do something simple like this:

import time

import pulsar

client = pulsar.Client("pulsar://localhost:6650")

reader = client.create_reader(
    "persistent://foo/bar/my-topic",
    start_message_id=pulsar.MessageId.latest,
)

reader.seek(int(time.time() - 1*60*60) * 1000)  # Seek to 1h ago.

for _ in range(5):
    start_time = time.time()
    reader.has_message_available()
    print(time.time() - start_time)

For both pulsar-client==3.4.0 and pulsar-client==3.5.0 the first has_message_available() call takes about a second, but after that I'm finding that each call takes about 5e-6 seconds for v3.4.0 (pretty quick), but about 0.2 seconds for v3.5.0 (a lot slower).

There were some bug fixes regarding has_message_available() and seeking that were introduced in v3.5.0 -- it's possible that the slowdown I'm running into might be related to that?