reduce memory footprint of Mojafos deployment

[tdaly61]

We want Mojafos to be simple and cheap to use over a good number of platforms to this end we want to reduce the memory footprint of he default deployment from the current (approx) 16GBs to as low as possible and ideally to 8GB.

If 8GB can be achieved then Mojafos can likely be run on a single Raspberry PI -4 with 8GB and though this might be a stretch goal it is one worth trying for.

Initial observation of deployment Mojafos (1xOHEE, 1xMifos and 1xPHEE instances) show that there are still some obvious things to target

• reduce number of MySQL databases to 1 (down from 2 currently)
• reduce number of kafka deployments to 1 (down from 2 currently)
• reduce number of redis instances deployments to 1 (down from 3 currently)
• reduce Java heap sizes (some are still Xms=1Gb)
• potentially reduce monitoring (ElasticSearch and Kibana and make optional if this is possible and not going to affect function)

Note: the approach here is to avoid adding to the plethora of existing helm charts and values.yaml files already available in the Mifos Repos but instead use and adopt the existing configure.py facility from vNext to adjust the memory params and values.yaml files at deploy-time so that they are specific and limited to Mojafos requirements.