[Meta][Metricbeat] - Collect even more additional Elasticsearch node metrics for enhanced dashboards

In order to improve stack monitoring, there needs to be some additional datapoints made available in a flattened view to enable visualizations and alerting.

Some of this data exists in the cluster_stats, but it being nested makes it very difficult to utilize.

Data that we would like to have either in the node_stats metricset (or a new metricset) is largely available by calling the _nodes API. Period for this might make sense to be different from existing node_stats as these items would not change often.

From the _nodes API

node.roles - this being stored in arrays, causes a single node to appear as multiple rows in a table visualization (one for each role), making it difficult to get a concise, single-line overview of a node. could this be as a single list instead?
node.attributes.availability_zone
node.attributes.region
node.instance_configuration - cloud only, but useful if available
node.version (i think this would be useful to track status of rolling updates)

node.os.cpu.processors: The number of logical CPU cores. This provides the denominator for calculating CPU load percentage, allowing for more effective alerting thresholds (e.g., load_avg_1m / processors > 0.8).

Thread Pool Statistics: Exposing the size and queue size for key thread pools is vital for diagnosing bottlenecks.
node.thread_pool.force_merge.size
node.thread_pool.force_merge.queue_size
node.thread_pool.get.size
node.thread_pool.get.queue_size
node.thread_pool.search.size
node.thread_pool.search.queue_size
node.thread_pool.write.size
node.thread_pool.write.queue_size
node.thread_pool.management.size
node.thread_pool.management.queue_size
node.thread_pool.esql_worker.size
node.thread_pool.esql_worker.queue_size
node.thread_pool.refresh.size
node.thread_pool.refresh.queue_size
node.thread_pool.snapshot.size
node.thread_pool.snapshot.queue_size

Quality-of-Life Improvement: Pre-calculated Percentages
As an additional, related enhancement, providing pre-calculated disk and memory usage percentages would be highly beneficial. While the raw used_in_bytes and total_in_bytes fields are available, exporting a calculated disk.used.pct and memory.used.pct field would simplify dashboard creation and alerting for all users.

Start by inspecting the Metricbeat node_stats metricset and existing cluster_stats data, then compare the available fields with the Elasticsearch _nodes API. Define which node roles, attributes, version, processor, thread-pool, disk, and memory values belong in the metricset, and verify that the resulting fields are flattened and suitable for dashboards and alerting.

Written by the indexing model from the issue text.