Improve config defaults for async secondary index MR jobs

Our defaults for the MR job that builds secondary indexes should be changed to better support larger tables.

• Increased the number of hbase client retries to 10
• Increased the hbase scanner time out, map reduce task timeout, hbase rpc timeout to 10 minutes instead of 1 min
• Added ability to control MR job concurrently running mapper tasks to 20 by default and a phoenix property phoenix.mr.concurrent.map.limit using which one can configure the same.