[REQ] Migrate enable per request authentification in new python codegen

Is your feature request related to a problem? Please describe.

Actually in current python authentification is managed in api configuration.
But when you instantiate a new configuration, a new thread poll is create with it, so each time you need to change your authentification key, you will create a thread poll which will pull down the performance of the application.

We could have a global configuration and change the authentification but it will not be thread safe.

Describe the solution you'd like

The solution was already developed in python-legacy but it has not be migrated.
As designed in this PR #6569
Overwrite the authentification in the request

[spacether]

@LeComptoirDesPharmacies is adding the missing auth information to your spec and regenerating the client an option?

Hi @spacether,

More details about our use case :
Idea is that we are doing some JWT forwarding in some of our services. We got the key from the received client request and reuse it against external services where client code is generated through OpenAPI codegen.
This JWT forward process is highly dynamic and multi-threaded.

About your suggestion :
Auth is already in the yaml spec file. Generated code allow to set "Authorization" header in the API client Configuration. However this configuration is shared/reused along all client API. This is not compliant with our multi-threaded context due to the fact that if a thread change the value of Auth in the Configuration instance, it could break the requests in other threads.

A workaround would be to instantiate an API Client / Configuration for every request coming on our server, before doing the JWT forwarding. However, this will create a very large amount of "useless" API client instance and each one will create a new pool of thread which is terrible for performances.

Idea of this issue/PR is to add capability to set distinct JWT value on every client API call, which will allow to get current-thread dependent value (thread-safe) and high dynamic.

Here is a PR proposal to add this feature in the current standard Python Codegen :
#11279

Yours faithuflly,
LCDP