Don't regenerate operation_id in MockResolver

[RobbeSneyders]

When the same operation was being resolved multiple times (eg. in different middleware), a new mock operation_id was generated every time, breaking routing for these mocked operations.

connexion.middleware.abstract.MissingOperation: Encountered unknown operation_id

This PR keeps track of the generated operation_ids and returns a previously generated id if the same operation is resolved.

CC: @julienschuermans

[coveralls]

coverage: 94.39% (-0.02%) from 94.408%
when pulling 4803cf1 on bugfix/mock-operation-id
into 3450a60 on main.

[Ruwann]

Thanks, left a minor (optional) comment.

Can we add a test for this? Doubting whether it is overkill to do this for the benefit we'll get from a test for the mock resolver but if only for the resolver, a test should be that hard to add?

[Ruwann]

In this branch, we know it exists, so we should be able to directly use access by key?

[RobbeSneyders]

Looking into adding a test makes me doubt if this is the right way to resolve this issue.

We are creating a new SwaggerOperation / OpenAPIOperation in every middleware (code), which might not be ideal. I'll check if we could handle this differently when I find the time and update this PR.