Capture usage telemetry on queries

We'll want to capture some telemetry on queries. Specifically, we'll want a histogram for the reporting interval (1h) for `name` - `[buckets]`:

* `query_response_time_ms` - `[10, 30, 50, 100, 200, 400, 800, 2000, +]`
* `query_file_total` - `[1, 2, 6, 12, 24, 140, 280, 432, 500, 1000, +]`

We'll also want a counter for the number of query requests that returned with an error due to the file limit being hit.

No files or tests are named. Start by locating the query request handling and the path that reports file-limit errors; identify where the one-hour response-time and file-count histograms, plus the file-limit error counter, should be recorded. Done means these metrics are emitted with the specified buckets and error condition.

Written by the indexing model from the issue text.