Reduce per plasma instance fixed memory overhead to support large number of indexes

Indexer storage mode used: plasma

Test which shows the failure:

./testrunner.py -i <ini> -p gsi_type=plasma -t rqg.test_rqg.RQGTests.test_rqg_concurrent_with_secondary,test_file_path=b/resources/rqg/simple_table_db/query_tests_using_templates/query_50000_aggregate_functions.txt.zip,database=simple_table_db,reset_database=True,concurreny_count=5,index_quota_percent=30,password=password,use_mysql=True,replicas=0,run_query_with_secondary=True,run_explain_with_hints=True,total_queries=2000,record_failure=True,check_covering_index=True,failure_record_path=/tmp,skip_cleanup=True

Leaving the cluster as it for debugging:
http://172.23.107.165:8091/

Attaching screenshot of RAM quota for indexer which was set deliberately to see if indexer goes into DGM mode.

This test creates indexes in batches using defer_build=true.
and issues explain queries against these indexes.

When some of the indexed are in ready state, I checked the index stats of a particular index and I do not see any data in data size and total items remaining area of Index stats.
Attaching screenshot of a particular index.
Also attaching logs for this particular run.

The test is passing and there is no failure seen .