Add pause/resume logic in and host health awareness in consumers

[Rahmeen14]

Problem statement

Consumers in rmqcpp keep consuming from a queue regardless of the health of the host they're running on. In environments where a host can become degraded (e.g. losing connectivity to a downstream dependency while still
connected to RabbitMQ), it's desirable for a consumer to stop taking messages off the broker until the host recovers, so that in-flight messages aren't processed on an unhealthy host and can instead be delivered to a healthy consumer.

There was previously no mechanism to pause/resume a consumer's delivery flow based on host health.

Proposed changes

This PR adds opt-in host-health awareness to consumers so they can stop pulling messages while their host is unhealthy and resume once it recovers. So messages aren't processed on a degraded host. It's driven by a periodic health check and enabled per-consumer through a new host-health configuration; when that config isn't set, behavior is completely unchanged. The change also adds metrics for health checks and consumer latency to make the behavior observable.

CI / build fixes (included in this branch)

• Rebuilt dockerfiles/amqpprox.Dockerfile on ubuntu:24.04 with Conan 2 (the previous conanio/clang9 base was no
  longer usable).
• Bumped cmake/conan versions accordingly.
• Integration test fix: force_close_connections parsed amqpprox's conn print output with a regex written for the old session format. amqpprox (Add connectionName field and structured log format for sessions amqpprox#120) changed it from <id>: vhost=<vhost> to id=<id> vhost=<vhost> …; the busted Docker cache surfaced this. Updated the regex to accept both formats, and extended the post-force-close assertion timeout to match test_connect_disconnect.

Testing

• Unit tests added/updated for HostHealthMonitor, ReceiveChannel pause/resume, ChannelMap, Connection, ConsumerConfig, and RabbitContextOptions.
• Integration test suite (rmqperftest / rmqapitests) passing, including the previously failing test_forceful_disconnect.

Notes for reviewers

• The pause/resume feature and the CI/test fixes are in separate commits for easier review.
• Host-health checking runs on the connection's event loop; the check implementation is intentionally pluggable/config-driven.