Pull subspace in - do not merge until point met

[calum-chamberlain]

@cjhopp has been working on a subspace branch, but it needs some refinement. Before we merge this in, the points below should be met:

• subspace detector functions;
• tests written for subspace detector (synthetic data and real data);
• test different multiplexing technique;
• optimize inner loop - cython?;
• tutorial;
• documentation.
• a full, efficient looping method for multiple detectors;
• plotting of the detector object (will only really be useful if not multiplexed);
• Add network coincidence method.
• test to confirm detections for multiplexed and non-multiplexed are the same.

- [ ] test for utils.clustering.min_sub_dimension

[calum-chamberlain]

Now operates in an object-oriented sense using the Detector class. Docs and tutorials written, just need to finalise tests.

[calum-chamberlain]

Non-multiplexed detection requires lagging, which requires delays to be tracked... Not sure how to do that.

[coveralls]

Coverage increased (+1.8%) to 77.881% when pulling 589afa8 on feature/subspace into 94da62a on develop.

[calum-chamberlain]

Everything is added, merge once green!

[coveralls]

Coverage increased (+2.0%) to 78.078% when pulling d2b3006 on feature/subspace into 94da62a on develop.

[coveralls]

Coverage increased (+2.0%) to 78.078% when pulling ec40725 on feature/subspace into 94da62a on develop.

[calum-chamberlain]

Issue in creation of subspace detector - different on different systems. I think this relates to the alignment of traces, will write an explicit test for this now and check.

[coveralls]

Coverage increased (+2.09%) to 78.198% when pulling b0dd3bb on feature/subspace into 94da62a on develop.