[ML] Factor out duplicated code in model library unit tests

[droberts195]

Many unit tests in the model library start off by creating a model, and there is a lot of duplicated code that does this, both within the suites for the different types of models and between the suites.

The affected files are:

CCountingModelTest.cc
CEventRateModelTest.cc
CEventRatePopulationModelTest.cc
CMetricModelTest.cc
CMetricPopulationModelTest.cc

For example there are:

• Large blocks of code that call addArrival
• Large blocks of type definitions
• Free functions within the different test files that help with model setup
• All these classes use a test fixture that has a protected resource monitor

There should be scope to refactor this to reduce duplication and verbosity. Some ideas are:

• Create a separate file containing a test fixture base class that all the model tests can share (in Boost.Test a test fixture is a class that every unit test that uses it inherits from, so can use its protected methods and data as a code reuse mechanism)
• Make some of the local setup helper functions methods of this base class so that they can be shared between the different suites of model tests
• Move commonly used type definitions into this shared test fixture - then they will be available to every test without needing to be redefined within it
• Add some methods that make it easier to add batches of arrivals
• Modernise the code using initializer lists rather than batches of push_back calls
• Add a test setup factory class to encapsulate some aspects of creating a model for unit testing

[edsavage]

This list will be used to check off each constituent piece of work as it is completed and updated with the corresponding PR number.

• Create model test fixture base class [ML] Refactor model lib unit tests #1499
• Move common type definitions to base class [ML] Refactor model lib unit tests #1499
• Move resource monitor instance to base class [ML] Refactor model lib unit tests #1499
• Modernize code where possible:
  • use initializer lists to populate containers where possible [ML] Modernize and tidy model test code where possible #1509
  • use range based for loops in preference to e.g. boost::size [ML] Replace boost::size in model unit tests #1517
• Refactor local helper functions to be members of base class [ML] Move model test helper functions to base class #1523
• Amalgamate the multiple addArrival methods [ML] Amalgamate addArrival methods #1536
• Add test setup factory class to create appropriate models [ML] Provide factory setup for creating models #1527