Remove log pollution from charm tracing in scenario tests

[mmkay]

Issue

In scenario test runs, a significant number of warnings (and some error logs) came from the fact that charm tracing was enabled even though there isn't anything charm traces can be sent to.

WARNING  tracing:charm_tracing.py:658 <class 'charm.GrafanaAgentMachineCharm'>._server_cert is None; sending traces over INSECURE connection.
WARNING  opentelemetry.trace:__init__.py:537 Overriding of current TracerProvider is not allowed
WARNING  opentelemetry.sdk.trace.export:__init__.py:213 Already shutdown, dropping span.
WARNING  opentelemetry.sdk.trace.export:__init__.py:213 Already shutdown, dropping span.
WARNING  opentelemetry.sdk.trace.export:__init__.py:213 Already shutdown, dropping span.
WARNING  opentelemetry.sdk.trace.export:__init__.py:213 Already shutdown, dropping span.
WARNING  opentelemetry.sdk.trace.export:__init__.py:213 Already shutdown, dropping span.
WARNING  opentelemetry.sdk.trace.export:__init__.py:213 Already shutdown, dropping span.

Solution

Disable charm tracing in all scenario tests as we aren't testing charm tracing there. We're testing tracing-related configuration in one set of tests, but this doesn't need charm tracing to be enabled.

Context

Testing Instructions

Run tox -e scenario and see the log pollution decreased significantly.

Upgrade Notes

[MichaelThamm]

Still seeing errors related to charm:charm.py:632 snap file for foo not found. It is likely not installed. Skipping.

But this PR fixes the spam of warning messages and will address the other issues in a different PR. Thanks!