Testing http-Specs with oas3 emitter

[skywing918]

1 . Verify that all http-specs cases can be converted to the OpenAPI 3 without any errors/warnings

describe("http-specs convert", () => {
  defineSpecTests(config);
});

2. Check information for a http-specs case.

it("authentication/api-key", async () => {
      const curr = {
        name: "authentication/api-key",
        fullPath: resolvePath(specsRoot, "authentication/api-key"),
      };
      const results = await openApiForFile(curr);
      const res = Object.values(results)[0] as any;
      const getThing = res.paths["/authentication/api-key/invalid"].get;
      strictEqual(getThing.operationId, "invalid");
    });

3. support 3.1.0

[skywing918]

@lirenhe pls review

[timotheeguerin]

why do we need to start the server

[wanlwanl]

@timotheeguerin, we use the response as ground truth:

          await validataDataWithSchema(response.body, invalidAuthSchemas);

[timotheeguerin]

I fail to see the real value in this PR. Do we really need this? You go and reimplement everything to validate pretty much nothing. This is a lot of wasted effort.

What is the motivation behind this?

[lirenhe]

The goal is to leverage the full test specs in Spector to verify various scenarios for the OAS3 emitter and display the results on the Spector dashboard for scenario coverage. This way, any changes in TypeSpec/Emitter can be quickly identified for potential breakings.

But I am not quite get the point in this PR about how we validate the result. Are we using snapshot for testing?

[timotheeguerin]

Fail to see why we need this