Use checksum to determine if the dataset has to be downloaded again

[albangossard]

Currently, we're downloading the dataset from the server (google drive) either if it doesn't exist locally, or if it's older than one day. This solution is not satisfying as the dataset will increase in size.
What could be implemented is to compare the local checksum to the one of the server. If the two doesn't match, then the dataset is downloaded again. If the user doesn't have an internet connection and the data exist locally, it should skip that step and not raise an error.
For more information, refer to https://github.com/ODINN-SciML/Sleipnir.jl/pull/97/files#r2056498136