longruns global seasonal cycle

[AlexisRenchon]

closes #913

We currently plot global maps for initial and final time steps (months)
of the long runs. In this commit, we add a global seasonal plot.
This plot takes the monthly global average of each output variables
and plot them as a line, the simulation is one year long so there is
12 data points.

TO DO:

1. Apply ocean mask:

var_no_ocean = ClimaAnalysis.apply_oceanmask(var)

Note: this sets values to 0.0 in the ocean. When calculating averages, it would drag toward 0. We need to replace with NaN or missing. Probably:

var_mask_missing = ClimaAnalysis.replace(var, 0.0 => missing)

2. Multiple lines for multiple years
   Currently longruns are one year, but they will be more soon. (can test 2 years in this PR). plot each year in light grey, and the average in thick black.

3. And of course, add axis labels, legend, etc.