Pre-run models for RAM issues

[PabloGGaray]

The notebooks friedman.ipynb and friedman_trees_alphas.ipynb need around 21 GB of RAM to run the models in that notebook. Maybe it could be useful to upload to the repository a pre-run model for these notebooks, to have the option of loading the models instead of running them when the user doesn't have enough memory.

[PabloGGaray]

Another option could be to use psutil to check the total RAM (based on this), with something like this:

from psutil import virtual_memory
mem = virtual_memory()
ram = mem.total/(1024.**3)

For friedman.ipynb

if ram > 20.0:
    num_covariables = ["5", "10", "100", "1000"]
elif ram < 20.0:
    num_covariables = ["5", "10", "100"]

And for friedman_trees_alphas.ipynb

if ram > 20.0:
    alphas = [0.1, 0.25, 0.5]
    idatas_at = {
        "10": {"0.1": {}, "0.25": {}, "0.5": {}},
        "20": {"0.1": {}, "0.25": {}, "0.5": {}},
        "50": {"0.1": {}, "0.25": {}, "0.5": {}},
        "100": {"0.1": {}, "0.25": {}, "0.5": {}},
        "200": {"0.1": {}, "0.25": {}, "0.5": {}},
    }
    trees = [10, 20, 50, 100, 200]
elif ram < 20.0:
    alphas = [0.1, 0.25, 0.5]
    idatas_at = {
        "10": {"0.1": {}, "0.25": {}, "0.5": {}},
        "20": {"0.1": {}, "0.25": {}, "0.5": {}},
        "50": {"0.1": {}, "0.25": {}, "0.5": {}},
    }
    trees = [10, 20, 50]