Test data bundled with repo to cover all permutations

[SamBryce-Smith]

Mainly to allow local testing of pipeline plumbing with locally valid paths.

The test data on cluster is handy but probably too big as still need to submit.

Should make:

• Tiny FASTQs (like 1/2 MB max), PE and SE
• FASTQs with 'multiple units' (i.e. same sample, reads split across multiple files)
• Sample tables and config files to match these specs (one to test PE, one to test SE, at least one sample has multiple units).

Ideally, should be able to run locally on cluster (on an interactive job) and should finish in 5-10 mins (test runs take >1hr atm which is a pain).

[SamBryce-Smith]

#32 is a perfect example of why we need this...

[SamBryce-Smith]

Partially fixed by #45 - need to add example sample tables with units & single-end