Update fastapi to 0.65.2 and dependencies

[davelopez]

This supersedes #12128 and upgrades fastapi and also its dependencies as some of them include security fixes:

• fastapi from 0.63.0 to 0.65.2 see changes here
• starlette from 0.13.6 to 0.14.2 see changes here
• pydantic from 1.7.3 to 1.8.2 see changes here

I tried running poetry update but for some reason, in my machine, it keeps running for several hours and seems to be stuck in a loop. In the end, I tried to update the dependencies manually, but maybe someone has more luck/experience with poetry to make it work.

How to test the changes?

• This is a refactoring of components with existing test coverage.

License

• I agree to license these contributions under Galaxy's current license.
• I agree to allow the Galaxy committers to license these and all my past contributions to the core galaxy codebase under the MIT license. If this condition is an issue, uncheck and just let us know why with an e-mail to galaxy-committers@lists.galaxyproject.org.

[davelopez]

Investigating the failed tests...

[davelopez]

Ok, apparently the only breaking change is that now clients that send requests with JSON payload must specify the Content-Type explicitly in the headers, see this comment

It looks like we do so in the POST but not in PUT or DELETE (or any other method that may accept body payload) in the GalaxyInteractorApi. In 3489225 there is a quick&dirty patch to fix the tests, but I guess it would be nice to add the as_json parameters to them as well and maybe even use True as default. Maybe in its own PR?

[bgruening]

Can you add a "todo" to the last commit? I guess this hack needs to go away sooner or later.

[davelopez]

I guess the timeout in the failed test is unrelated since I only added the TODO comments in the last commit.

[mvdbeek]

I guess the timeout in the failed test is unrelated since I only added the TODO comments in the last commit.

Yeah, timeout is unrelated. The preceding tests are uploading datasets but are not waiting until the upload is finished, so the job queues up and doesn't finish within 60 seconds. You can see it in the logs, job 21 is the job we're waiting for, while job 17/18/19/20 are still running