penumbra: fix metrics by bumping helper crates

[erwanor]

In #3745, we bumped the metrics crate to use 0.22.

However, we missed a couple of its helper crates such as:

• metrics-utils
• metrics-exporter-prometheus
• metrics-tracing-context

This is problematic, and apparently frequent cause of breakage, because ones need to use the same version of metrics across crates that register recorders in order for it to work.

One way to detect this mismatch is running this command on the workspace directory:

cargo tree -i metrics

Later, we discovered that - unfortunately - the [email protected] crate had pinned of its dependency ([email protected]) in order to enforce its MSRV policy (!), effectively locking us - and astria - out of completing our migration to 22.

Fortunately, after some outreach a new release removing the pinned dependency was cut and we can now complete. This PR bumps the workspace to use appropriate crates. It should fix the metrics breakage.

[conorsch]

Confirmed via local node, joined to preview, that this change restores metrics emission for pd:

❯ curl -s http://localhost:9000/metrics  | tail
# TYPE penumbra_dex_path_search_duration_seconds histogram
penumbra_dex_path_search_duration_seconds_bucket{le="0.00001"} 0
penumbra_dex_path_search_duration_seconds_bucket{le="0.0001"} 1270
penumbra_dex_path_search_duration_seconds_bucket{le="0.001"} 1762
penumbra_dex_path_search_duration_seconds_bucket{le="0.01"} 1762
penumbra_dex_path_search_duration_seconds_bucket{le="0.1"} 1762
penumbra_dex_path_search_duration_seconds_bucket{le="+Inf"} 1762
penumbra_dex_path_search_duration_seconds_sum 0.162245319
penumbra_dex_path_search_duration_seconds_count 1762

Closes #3780.