Choose a consistent file structure for integration test data

[dstansby]

In #101 I'm trying to simplify the integration tests so it's really easy to add new tests. I've noticed that there is not a consistent file structure in the test files that have been uploaded. For the low numbered tests (01, 02, 03) the structure is a single directory containing

• a single input file named pstd_input_file.m
• one or more input data files named pstd_{run_type}_input.mat.
  run_type can be any string.
• corresponding reference output data files named pstd_{run_type}_reference_output.mat.
  For every input file there is a corresponding output file with the same run_type in the filename.

But in the high nubmered tests (09, 10) there are "in" and "out" directories. To make the test infastructure as simple as possible, it would be good to choose, document, and stick to a file structure for the integration test files.

[samcunliffe]

Do you have a preference?

Peter uses directories in/ and out/ in his tarballs so

• [Pro] it's in keeping with his workflow & scripts.
• [Con] *_reference_output.mat is arguably more explicit.

[dstansby]

I don't personally have a preference and think either is fine, so I'd say go with what Peter prefers since he'll be using the software after the ARC project finishes

[samcunliffe]

Solved by @willGraham01 (quite a while ago now... I've lost the PR)

[willGraham01]

This feature comes in with #236, however reaches it's final (and present solution state) in #244.

The config_XX.yaml files explicitly provide the full path to necessary input/output/reference files for each test.