Fix issue #84

[lohedges]

This PR closes #84. We now handle missing values in AMBER output files by extending the regex to parse records such as PRESS =********. In all cases, any missing values are interpreted as None. I've modified one of the files used in the AMBER parsing tests to contain a missing value in order to confirm that things work as expected. Currently this code is only applied to the Exscientia Sandpit, where the records are being used. (They are saved to disk after a run.) I'll move this into the core code when I port over the AMBER FEP code.

I've also updated the plotting code to handle None values in time-series data with units. When None is present we filter the both the x and y data (and any associated errors) to remove data corresponding to the None indices in both sets, i.e. if None is present in x, then we remove this value from x and the same index from y. I've not updated the logic for raw float values since this isn't used internally by BioSimSpace, nor the code for plotting contours for 3D data.

I've also removed some trailing whitespace that was introduced in PR #107.

• I confirm that I have merged the latest version of devel into this branch before issuing this pull request (e.g. by running git pull origin devel): [y]
• I confirm that I have permission to release this code under the GPL3 license: [y]

Suggested reviewers:

@chryswoods

[chryswoods]

This looks great.