Modularisation of FMTM Splitter

[spwoodcock]

Is your feature request related to a problem? Please describe.

• Ivan actually did this a while ago, but we pulled out the changes he did in modules and dumped them in a single file for simplicity.
• This was done to reduce the overhead of multiple SQL queries on the databases - we simply have one connection, one cursor, one query.
• However it has come to light it might be best to modularise this again.

Describe the solution you'd like

• The fmtm_algorithm.sql file should be divided into modules, described in How The FMTM Splitter Algorithm Works fmtm-splitter#61
• From Python each module can be called via a new cursor.
• The output from one cursor will be used as the input for the next SQL statement.

Additional considerations

• The main advantage here is that we can add parameters to each stage, allowing us to tweak the output as needed.
• It is quite clear that splitting is unique in each new area we are presented with - we need flexibility to generate the optimal split for a project.