first draft of greta_sitrep() for loading greta and establishing if t…

[njtierney]

…he software is ready to use

See #441 for linked issue.

It currently looks something like this:

Although I'm not 100% sure why "loaded Tensorflow 1.14.0" appears in the middle of the text.

I'm not 100% if it should initialise Python, and I need to test out how it fails in specific situations and what sort of instructions appear. For example, it should be OK if users don't want to use a greta conda env.

To do that I'll need to write some mocking tests, I think?

@goldingn do you have any thoughts on how this is presented and if this is useful? Do you think that a separate, load_greta_deps() or similar function should be used to load and establish Python (instead of needing to run greta code to trigger that step)

[goldingn]

This looks helpful! I think the initialisation is good. And necessary to make sure everything works.

Should the version numbers for TF and TFP be reported when saying what is available, to reiterate the requuirement and what is checked?

[njtierney]

Ok great, I'll leave the initialisation in then.

Yup, agree! Having thinking about this a bit more this past week I think a few more helpers on what to do if each part isn't installed are there. So overall:

• [x] report the version numbers of each of TF, TFP, Python
• [x] if there is an error, report where things do not match (e.g., TF not found, or TF installed is not the right version)
• [x] provide information on what to do in the case that each of these things isn't found (most likely that would be to just run install_greta_deps()
• [x] add mocking tests for when the software is not available
• [x] add mocking tests for when the software versions are different

[njtierney]

This is what it currently looks like:

Note this also removes the errant "Loaded Tensorflow 1.14.0" message as well.

[njtierney]

reduced exclamation marks

[njtierney]

Added mocking tests for different versions of software and software availability