Pass Registry to Counter and Gauge constructors

[nss10]

Description about what this pull request does.

• This PR updates the increment_counter method in cdispyutils/metrics.py to register counters with the instance’s CollectorRegistry instead of the default global registry.

Why this change is needed:

• Without specifying registry=self._registry, the Counter is registered in the default global registry provided by prometheus_client.
• When the same process (e.g., during tests, multi-threaded code, or repeated initialization in a long-running service) tries to create a counter with the same name in the default registry, Prometheus raises the following error:

ValueError: Duplicated timeseries in CollectorRegistry

• More info : here
• By passing registry=self._registry, each MetricsHandler instance maintains its own isolated CollectorRegistry. This prevents accidental clashes with other metrics in the default registry.

• Why the error never happened in production but occurred in unit tests
  1. Production services usually initialize metrics once
     • In a live deployment, the process starts, creates its metrics in the default registry, and keeps running. No duplicate registration occurs unless the process restarts.
  2. Unit tests often create multiple instances in the same process
     • Many tests spin up fresh MetricsHandler objects without restarting Python.
     • If each test case creates counters with the same name in the global registry, duplicates are inevitable.

Please make sure to follow the DEV guidelines before asking for review.

New Features

Breaking Changes

Bug Fixes

• Fix: Prevent duplicated timeseries in Prometheus metrics by using instance-specific registry

Improvements

Dependency updates

Deployment changes

[github-actions]

The style in this PR agrees with black. ✔️

This formatting comment was generated automatically by a script in uc-cdis/wool.

[george42-ctds]

Looks good!

[george42-ctds]

Looks good!