replicated_fast_count_manager.cpp contains numerous server status metrics:

// Boolean flag indicating whether or not the fast count background thread is currently running.
Atomic<bool> isRunning

{false}

;

// Flushes persist fast count information to the oplog and occur during checkpointing,
// shutdown, step down, etc. The total number of flush attempts = flushSuccessCount +
// flushFailureCount.
Atomic<int64_t> flushSuccessCount

{0};
Atomic<int64_t> flushFailureCount{0}

;
Atomic<int64_t> flushTimeMsMin

{std::numeric_limits<int64_t>::max()}

;
Atomic<int64_t> flushTimeMsMax

{0};
Atomic<int64_t> flushTimeMsTotal{0}

;
// Aggregate metrics for the min/max number of documents inserted or updated during one flush.
Atomic<int64_t> flushedDocsMin

{std::numeric_limits<int>::max()}

;
Atomic<int64_t> flushedDocsMax

{0};
// The total number of documents written during flushes. Used to compute the average flush size.
Atomic<int64_t> flushedDocsTotal{0}

;

// The number of times an empty diff is found when writing an update to the replicated fast
// count collection.
Atomic<int64_t> emptyUpdateCount

{0};

// The number of inserts into a new record for storing size and count data.
Atomic<int64_t> insertCount{0}

;
// The number of update to an existing record storing size and count data.
Atomic<int64_t> updateCount

{0};
// The total time spent writing metadata to the replicated fast count collection.
// writeTimeMsTotal / flushTimeMsTotal = the proportion of iteration time writing dirty
// metadata.
Atomic<int64_t> writeTimeMsTotal{0}

;

For each metric, use the corresponding mongo::otel::metrics metric.

In generateSection(), there are conditions for reporting the values of min and max. There is also a
recordFlush() member function. Preserve these implementation details by authoring a class or struct
that wraps the otel metrics and provides an API for modifying them. Define this class in
replicated_fast_count_metrics.

{h, cpp}

.

Write unit tests in replicated_fast_count_metrics_test.cpp using the explanation provided in
otel/metrics/README.md. Evaluate whether or not the replicated_fast_count_server_status.js still
provides helpful test coverage.

As part of this change, check if there is an OTel ObservableMutex API for the _metadataMutex.

Retain all documentation and metric call sites.