Bugfix providers even further

[henrik-wolf]

I fixed a bunch of problems and inconsistencies I came across when playing around with different tile providers. Here they are:
In the NASAGIBS file:

• Empty constructor for the NASAGIBSTimeseries failed due to the default keyword not being in the list of valid ones.
• Fixed and extended docstring of NASAGIBSTimeseries (still, somehow printing about 800 lines of variantlist makes the docstring basically useless in the REPL...)
• Added NASAGIBSTimeseries to PROVIDER_DICT

in the providers file:

• set keyword in the docstring to always be lowercase, since the keywords of the functions are as well
• moved check for existence of variant one line up, otherwise we get the keyword of the Provider to show up as a possible variant in the error message
• split creation of functions with default arguments and keyword arguments in two steps, because there seems to be some strange behaviour with the methods function (see methods(f) not showing correct signatures for functions with default arguments JuliaLang/julia#49028)

Tests:

• reshuffled tests a bit to contain all Providers in the order they appear in the json

Readme:

• fixed the readme to show all provider functions and how to get urls as well as the PROVIDER_DICT

[rafaqz]

Thanks for all the fixes! There was a reason the NASAGIBBSTimeseries was not included in the readme or the providers dict... (it was so buggy)

Do you want to be a maintainer here? This package was hacked together very quickly while we put together Tyler.jl. It could do with some more ongoing care, but I'm unlikely to find time for it.

[yeesian]

(optional) Can you include a comment here to explain this line? Also, github's review UI complains when the file does not end with a newline

[henrik-wolf]

Yes. I will do that. This is just adding the nasagibstimeseries to the main dict of providers (as you probably know...). I will also add a new line.

[rafaqz]

@SuperGrobi might need to get out an old 1.6 version and see whats causing that failure. Probably one of the line you have added like the nada gibbs tests?

Something is wrong with replace on line 89, although the method signature looks fine to me.

[rafaqz]

Sorry I just realised this was against my other PR that I totally forgot existed, and not master