Add documentation

[QuentinMaz]

Hi,

I worked on a new way of assembling reference models with virtualizarr, as well as pre-processing raw .csv tag/biologging data.
I have since used this work and I think that my implementations can be beneficial for the community.
Consequently, I turned my code into two how-to guides (one about reference model creation, the other about raw biologging data) and added them in the documentation (under the section Data preparation).
I tried to make them didactic, but please note the following caveats:

• They both require access to the GFTS's S3 bucket (as all the other notebooks though).
• Creating a reference model requires credentials for copying data with rclone to the GFTS's bucket from Mercator Ocean. My goal was to best describe a practical use case, from finding the model reference to the final assembled dataset... But for sure a future improvement would be to assemble a dummy reference model without copying data to the bucket (and making the notebook way less computation intensive!). To account for this, I structured the notebook in two parts: the general methodology (with "template commands" to adapt), and a "real world" illustration of them to a similar use case I used the implementation for.

Also, related to the point above, I could not re-check the notebook about reference models (since I don't have the rclone credentials). But I checked the other notebook, and built the documentation locally.

[annefou]

Nice. Just a question: would it make sense to split the "data preparation" section? It looks to me that we now have much more than data preparation.
We could do it in another PR; what do you think?

[QuentinMaz]

Thanks for the answer!
As discussed in-person, I will update the structure of the documentation to distinguish the data processing content from the one about various examples of computation and visualization.