Analyze memory footprint increase at Hivemind sync

hive sync testing uncovered some memory leak / uncontrolled allocation increase in the python process.

• Identify cause of such increase.

Since initial analysis has been compleded, it shown that Posts._ids collection is a guilty, it should be eliminated. Looks it is used at:

• Communities
• Payments

We should create mock testcases to cover above usecases and eliminate this cache by adapting INSERT/UPDATE queries executed on database by taking original author/permlink pair instead of retrieved from cache post-id.