Add unit and integration tests for monitoring module

[gimballock]

Add test coverage for the monitoring/metrics system at both the unit and integration level.

Changes

Commit 1: Unit tests for monitoring data layer and HTTP API (1e45c73)

Unit tests for the previously untested monitoring module in stratum-apps/src/monitoring/:

• client.rs: Sv2ClientInfo accessors, to_metadata(), Sv2ClientsMonitoring default impls (get_sv2_client_by_id, get_sv2_clients_summary)
• server.rs: ServerInfo accessors with Option hashrate handling, ServerMonitoring default impl (get_server_summary)
• sv1.rs: Sv1ClientsMonitoring default impls (get_sv1_client_by_id, get_sv1_clients_summary including None-hashrate filtering)
• prometheus_metrics.rs: Conditional metric registration per role type, Prometheus text encoding round-trip
• http_server.rs: Pagination unit tests (effective_limit clamping, paginate edge cases), HTTP endpoint integration tests using tower::oneshot against mock monitoring sources for all 9 API routes and /metrics

Adds tower, http-body-util, hyper as dev-dependencies for the axum handler tests.

Commit 2: Integration test infrastructure and metrics assertions (1a6ef0e)

• Config setters (pool/config.rs, translator/config.rs, jd-client/config.rs): Add set_monitoring_address() and set_monitoring_cache_refresh_secs() so tests can enable monitoring on dynamic ports with a fast 1-second cache refresh.

• Test helpers (integration-tests/lib/mod.rs): Add start_pool_with_monitoring(), start_jdc_with_monitoring(), start_sv2_translator_with_monitoring() wrappers that configure monitoring and return the monitoring address.

• Assertion helpers (integration-tests/lib/metrics_assert.rs): Async helpers (fetch_metrics, fetch_api, assert_api_health, assert_uptime, assert_metric_gte, assert_metric_eq, assert_metric_present, assert_metric_not_present) using spawn_blocking internally for the synchronous minreq HTTP calls. 11 unit tests for the parsing/assertion logic.

• Metrics assertions augment 9 existing integration test functions across 8 test files. Each test waits 2 seconds for the SnapshotCache to refresh, then validates topology-appropriate metrics:

Test	Metrics validated
sv2_mining_device_and_pool_success	Pool: clients >= 1, uptime, health OK
pool_standard_channel_receives_share	Pool: clients >= 1, shares accepted >= 1, uptime, health OK
translate_sv1_to_sv2_successfully	Pool: clients >= 1, no server channels. tProxy: server channels present, SV1 clients >= 1, no SV2 clients, server shares accepted
test_extension_negotiation_with_tlv_in_submit_shares	Pool: clients >= 1, uptime, health OK
jd_non_aggregated_tproxy_integration	tProxy: server channels, SV1 clients >= 1. JDC: server channels, no SV2 clients. Pool: clients >= 1
jd_aggregated_tproxy_integration	Pool: clients >= 1. tProxy: exactly 1 server channel (aggregated)
jdc_submit_shares_success	Pool: clients >= 1, uptime, health OK
propagated_from_jdc_to_tp	Pool: clients >= 1, shares accepted >= 1
propagated_from_jds_to_tp	Pool: clients >= 1, shares accepted >= 1

Technical notes

• Metric helpers use tokio::task::spawn_blocking to wrap synchronous minreq HTTP calls, preventing deadlocks on the single-threaded tokio test runtime
• Tests use #[tokio::test] (current_thread) rather than multi_thread, because FallbackHandler::drop() in fallback_coordinator.rs panics if done() was never called — aborting translator/JDC tasks on a multi_thread worker triggers a double-panic SIGABRT
• Cache refresh is set to 1 second in tests (vs 15s default), keeping the per-test overhead to ~2 seconds

[gimballock]

I added some additional tests, let me know if you see any areas that still need more coverage

[plebhash]

PR title and description mention integration tests, but I don't see any on the diffs?

[gimballock]

Updated the integration tests to validate metric invariant related to their basic detecting ability: when connected channel counts should be non-zero, uptime should be positive, share count should be positive if we have seen a share, etc.