fix: overhaul reports

[Theodus]

The primary goal of this PR is to increase the amount of concurrent client requests the gateway can handle before p99 latencies start to go up above the latency of indexers responding to the client requests. This happens to align with cleaning up some of the gnarliest tech debt in the gateway that we've accumulated for backwards compatibility for data science. On my local testbench, these changes more than double the amount of concurrent client requests the gateway can handle.

The biggest changes are:

• use a reporting actor to limit kafka message sends to at most one task at any given time
• overhaul the client_query module to treat logging, metrics, feedback, kafka, etc. more clearly & deliberately

External impact:

• The old Client query result & Indexer attempt logs we kept around for internal data science pipelines are gone. Identical field names & values have been maintained in the corresponding kafka outputs. Though some unused fields like status_code and budget have been removed entirely.

[LNSD]

LGTM ✅

[LNSD]

I agree with the changes made to the client_query.rs file, but we could have waited to make the structural ones after merging the new network resolution module. I get that optimizing how we report the information is critical for the performance. But next time, let's try to keep changes scoped so we don't rework things over and over unnecessarily 😞