Add the option to save all temperature profiles in a climate run

[aditya-sengupta]

Closes #213.

• Currently demonstrates the feature in the 12a notebook, I can move this to somewhere else in docs. The only change other than the feature demo is I'm using a relative path to picaso/data/... on my machine; I could change this back but am not sure how to make the path work as it was. Is there a project convention to make paths starting with /data work, such as having /path/to/picaso in sys.path?
• Previously, all_profiles flattened the array of temperature profiles so that it was a 1D array of length (niter * npoints,). Changed this to be a 2D array (niter, npoints) and changed append calls to vstack to accommodate this. This means we can read the temperature profile at each iteration from the HDF5 file without needing to calculate indices for where each one starts.

[natashabatalha]

Awesome!!!

@aditya-sengupta could you do me a favor and submit the PR to dev, not master! Sorry I didn't mention that before! You should be able to edit this existing PR's base merge to dev easily

Also great questions about the paths. Currently the docs are built on a machine at nasa. I run all the notebooks during the build to ensure proper functionality of all parts before it goes live. Usually for file creation during a tutorial I just have it created in the same working directory for simplicity, then add the filetype to .gitignore. I think h5 files should already be ignored anyways.

[aditya-sengupta]

Noticed the 1D -> 2D change broke the convergence animation so added a fix for that

[aditya-sengupta]

Now saves to HDF5 whenever all_profiles is written to and returned from climate, instead of just at the end.

Currently write_all_profiles takes in the full set of T(P) profiles from the whole run and writes it to file from scratch, which means it's redoing writes from previous iterations. This is inefficient but to fix it we'd have to keep track of the overall iteration number as far as profile-saving is concerned, which gives us another thing we need to keep track of in climate. So this could be fixed but it'd be much easier to do once we have the climate call-stack diagram so I'm leaving it like this for now.