Rambaud Pierrick

Yeah sorry i was super tired of refactoring when I did `batch` module so I started being lazy with some mechanisms. I'll reimplement the `namePattern` one ASAP.

> if the collection had an intense computation it might fail due to memory issues. Instead, I propose to loop over the system:index property, which by definition is always unique...

is it opt-in ? Like can project relying on `pip` will continue to work ?

It will be necessary for my use case so i'll definitely look into it. I think the easiest way is to make num_regression independant from its dataframe counterpart ence recoding...

Actually I though about another solution that i already applied in my pytest-regressions extention for GEE object: [`pytest-gee`](https://pytest-gee.readthedocs.io/en/latest/): What if instead i was adding a "prescision" mechanism in data_regression ?...

I'm also a bit confused by this comment becasue to me prescision from numpy is absolute as well.

Usualy I have 2 builds on RDT: "stable" and "latest" if your link is not refereing any specific branch i.e. https://ipyvuetify.readthedocs.io/ then it will point to the default stable. If...

master is already relying on pydata-sphinx-theme but readthedocs latest is still pointing to rdt-sphinx-theme (default). So for a reason I cannot explain (as visitor cannot see the RDT project parameters)...

now I get your point, and then yes rediraffe is the one to go (I think they use it in matplotlib). should I code it in a following up ?

Would you want me to fill the different sections in this PR? I wanted to keep the scope of my PR reduced (as you suggested) so until this one is...