Document/publish our arrow schemas

[jleibs]

• The intent of Python API to get dataframes out of an RRD #6811 is to expose data to users as structured arrow data rather than using the Rerun-object-model.
• The intent of Stop using arrow unions #6388 is to make that arrow data cleaner and easier to work with.

However, so far the arrow data has been an implicit implementation detail sitting behind our object-APIs. To make this data accessible we need to be able to show users how to work with it.

Unfortunately, there is no standard textual representation of an arrow schema. Although we have our own fbs files, this is a Rerun-specific implementation detail that is not necessarily familiar to those attempting to cross-reference with arrow documentation.

For each Rerun datatype, we should generate a doc page specifically focused on the arrow schema:

• The terms used should as closely as possible matches the official terms from the arrow documentation for the native libraries.
• We should include a generated code snippet in each of the 3 languages that produces the corresponding arrow datatype. Note that we already generate this code in our serializers. We just need to emit that same code into a block in the doc pages.

We also need to formalize the documentation around any additional meta-data required for interpretation as a component or inclusion into a chunk.