Update DynamoDB TokenStorage example + mypy config

[sirosen]

In order to make the DynamoDB token storage example pass mypy linting and
use the current paradigm for TokenStorage, it needs:

• missing annotations (easy)
• retrofitting of existing functions (easy)
• mypy overrides to ignore boto3 types (easy, but note below)
• an implementation of get_token_data_by_resource_server (see note!)

For mypy, I have decided to move to a dedicated mypy.ini config in the
docs/ directory. With the config growing, it didn't make sense to keep
it embedded in the repo root pyproject.toml . That also lets us set
different top-level behavioral flags in the config.

For get_token_data_by_resource_server, I decided to implement it, in
the simplest possible way using a sequential scan, but no pagination.
It also prints a very annoying warning (printing, not emitting a warning,
so that warning filters are not applicable), to make sure that anyone
who picks it up and runs it verbatim gets some signal that they ought to
think about it and possibly make some changes.
The associated doc notes that you could make this method raise an error
and it would mostly work, but that some SDK features would fail if you
do so.

[derek-globus]

My only real note is around the term Adapter.

The old interface was called StorageAdapater, the new one is called TokenStorage. In my opinion, child classes shouldn't be referred to as "adapaters" in class name or text references to avoid confusion between the two versions.

[derek-globus]

Suggested change

	Unlike the builtin adapters for JSON and sqlite, enumerating tokens in storage
	is not a desirable operation.
	Unlike the builtin ones, enumerating tokens in a DynamoDB table-backed storage
	is not desirable.

Mostly a recommendation to avoid the word "adapter" here & in the rest of the document.

[sirosen]

I made a slight alteration in this when I applied, but spiritually it's the same. LMK if you want it tweaked more. 👍