PHOENIX-7704: Integrate HBase's new Scan Latency Metrics

[sanjeet006py]

JIRA: PHOENIX-7704

• This PR is consuming HBASE-29398, HBASE-28302, HBASE-29494 and HBASE-29626.

New metrics being added to Phoenix to be reported in query level read metrics:

• frt: Time spent in reading a HBase block from HDFS/filesystem.
• brff: Number of bytes read from HDFS to read the HBase Block. This will be case when the HBase block is not found in block cache or the block in block cache is corrupt.
• brfm: Number of bytes read from memstore. Here the size of each cell read from memstore is summed.
• brfc: Number of bytes read from block cache. This is the size of the block and the size computations are done in same manner as if the block was read from filesystem only. Thus, helps with apples-to-apples comparison of bytes read from FS vs block cache.
• broc: Number of block read operations. If the HBase block was read from filesystem then this metric is also incremented by 1 per HBase block read. This along with brff can give better idea of overhead coming from block read when bytes read are same while HBase blocks read have increased.
• rsp: Time spent in RPC scan processing. This is the time spent by scan RPC call scanning the rows by RSRpcServices#scan() method.
• rsqw: Time spent in RPC scan queue wait. Time spent by a scan RPC call in server side call queue.

[sanjeet006py]

PR validation build: https://ci-hadoop.apache.org/job/Phoenix/job/Phoenix-PreCommit-GitHub-PR/job/PR-2329/

[sanjeet006py]

The most recent commit changed only PhoenixMetricsIT but two tests failed in ParallelConnectionsIT so, clearly those are flappers.