Demo8 extremes update

[lee1043]

• Reduced number of lines (in cell [3])
• Clean up notebook
• Result from latest rerun

[lee1043]

@acordonez Thank you for resolving the issues for the extreme PR. I was able to make Demo8 notebook run it through.

I opened this PR for a few edits in the Demo8 notebook and also for some questions.

1. Compared to the original notebook, I see the numbers in JSON are slightly different, and the plot colorbar range is different (in the original 0 to 20, while in my notebook, 0 to 30). Do you consider they are close enough?

2. There are a lot of warning messages, curious if there is a way to reduce or mute them if they are not critical.

[acordonez]

@acordonez Thank you for resolving the issues for the extreme PR. I was able to make Demo8 notebook run it through.

I opened this PR for a few edits in the Demo8 notebook and also for some questions.

1. Compared to the [original notebook](https://github.com/PCMDI/pcmdi_metrics/blob/534_ao_extremes/doc/jupyter/Demo/Demo_8_extremes.ipynb), I see the numbers in JSON are slightly different, and the plot colorbar range is different (in the original 0 to 20, while in my notebook, 0 to 30). Do you consider they are close enough?

2. There are a lot of warning messages, curious if there is a way to reduce or mute them if they are not critical.

Regarding 1., the figure looks pretty similar to me once you account for the change in the colorbar. Unfortunately I don't think the various fits and solvers used to get the return values are guaranteed to produce stable results, but I'll look into this some more.

[acordonez]

@acordonez Thank you for resolving the issues for the extreme PR. I was able to make Demo8 notebook run it through.

I opened this PR for a few edits in the Demo8 notebook and also for some questions.

1. Compared to the [original notebook](https://github.com/PCMDI/pcmdi_metrics/blob/534_ao_extremes/doc/jupyter/Demo/Demo_8_extremes.ipynb), I see the numbers in JSON are slightly different, and the plot colorbar range is different (in the original 0 to 20, while in my notebook, 0 to 30). Do you consider they are close enough?

2. There are a lot of warning messages, curious if there is a way to reduce or mute them if they are not critical.

Regarding 2., I've tried a couple of ways to suppress warnings in the Jupyter notebook but have not been successful. I've tried the warnings module and the %%capture cell magic, but in both cases I still see all the warnings.

Warnings:

%%capture:

I'm reluctant to turn off all the warnings within the driver itself, as they are flagging real instances where the GEV fit is failing for one reason or another. Do you have any other suggestions?

[acordonez]

Regarding Q2, I've also tried redirecting output to a text file, but still get the warnings live:

[acordonez]

Here's the return value results I got for the first test run in the Demo, and I'm getting slightly different results from you and from my original run - not much matches after the decimal point. It wasn't unusual for me to get different results when I was testing the code on single timeseries, so I'm not surprised.

{
  "GISS-E2-H": {
    "r6i1p1": {
      "Rx1day": {
        "land": {
          "mean": {
            "ANN": 13.624667116247734,
            "DJF": 9.135238348527835,
            "JJA": 9.695412307925825,
            "MAM": 9.4947603968054,
            "SON": 10.085489430018676
          },
          "std_xy": {
            "ANN": 5.84894894720064,
            "DJF": 6.01362496906134,
            "JJA": 5.727293357791019,
            "MAM": 5.823173294700789,
            "SON": 4.879856373306809
          }
        }
      },
      "Rx5day": {
        "land": {
          "mean": {
            "ANN": 8.612030197868712,
            "DJF": 5.639787497943132,
            "JJA": 6.327905351825952,
            "MAM": 5.881106639945075,
            "SON": 6.422801996975585
          },
          "std_xy": {
            "ANN": 4.350543642971108,
            "DJF": 4.133318652361083,
            "JJA": 4.293765096254575,
            "MAM": 3.970500853809765,
            "SON": 3.6944120761563717
          }
        }
      }
    }
  }
}

[lee1043]

@acordonez thank you for the comments. I think we can be confident merging this if differences are only at decimal point level.