Document alternatives to API

[roed314]

There are three reasonable ways to access large amounts of data from the LMFDB:

1. Using the download functionality implemented in Download columns #5702, which lets you download search results with custom columns in different formats; limited to roughly 100MB of data per download
2. Using the API; doesn't allow an offset larger than 10,000
3. Using a direct connection to the devmirror.lmfdb.xyz postgres server, either via psycodict through LMFDB or LMFDB-lite, or via any other PostgreSQL front end.

We should decide what guidelines we should offer users on which method to choose and then document them (probably at the top of the API).

[AndrewVSutherland]

I like this suggestion, and I might even suggest making the Database link on the sidebar take you to a shorter/friendlier page that just explains these 3 options, with (2) including a link to the API page that is currently linked to.

I think the order you listed the options above is correct, and users should be directed to (1) first, which I would expect to cover most casual needs. The one use case where (2) might be preferred is in a situation where one wants to dynamically access the LMFDB in a simple automated way. (3) is for everything else (it requires more effort and sophistication on the part of the user but will let them do whatever they want).