Brig Polysemy: Galley RPC effect

[isovector]

This PR moves the majority of brig-to-galley's RPCs into a new GalleyProvider effect. In addition, it introduces two implementation-detail effects, ServiceRPC for routing RPCs to correct services, and RPC for actually sending the RPCs.

The vast majority of remaining MonadMask and MonadUnliftIO usages in brig correspond to RPCs. After doing the same treatment to cargohold and gundeck, it should hopefully be trivial to remove the MonadIO instance for AppT, at which point brig will be fully polysemized.

Checklist

• The PR Title explains the impact of the change.
• The PR description provides context as to why the change should occur and what the code contributes to that effect. This could also be a link to a JIRA ticket or a Github issue, if there is one.
• I updated changelog.d subsections with one or more entries with the following bits of information (details):

[mdimjasevic]

I haven't looked into details, but you have a handful of conflicting files with the upstream develop branch.

[isovector]

@mdimjasevic should be resolved now

[mdimjasevic]

This looks great and I like how you defined the GalleyProvider interpreters! I will rebase my PR #2450 on top of this once it gets merged because my approach was direct (without the RPC and ServiceRPC effects) and in the PR I didn't turn into an effect everything Galley in the Brig.IO.Intra, but only what I needed to get the code compiling.

You have a TODO or two and one undefined function. Can you address them before we merge the PR?

[fisx]

potentially unpleasant question: would it be possible to skip this effect altogether, given we're planning to merge brig and galley soon, and it'll all just be function calls?

[mdimjasevic]

@isovector , the PR fails with this in CI:

      TTL / Overrides

        increase to unlimited:                                                                                                                FAIL (0.38s)

          Error message: expected: FeatureTTLSeconds 2

           but got: FeatureTTLSeconds 1

          

          CallStack (from HasCallStack):

            assertFailure, called at ./Test/Tasty/HUnit/Orig.hs:86:32 in tasty-hunit-0.10.0.3-106eb1348444e345c4687ea268fe81c1fc84c38f18220dafc575b936429bc95c:Test.Tasty.HUnit.Orig

            assertEqual, called at ./Test/Tasty/HUnit/Orig.hs:108:23 in tasty-hunit-0.10.0.3-106eb1348444e345c4687ea268fe81c1fc84c38f18220dafc575b936429bc95c:Test.Tasty.HUnit.Orig

            @?=, called at test/integration/API/Teams/Feature.hs:504:18 in main:API.Teams.Feature

            getFeatureConfig, called at test/integration/API/Teams/Feature.hs:553:3 in main:API.Teams.Feature

            testSimpleFlagTTLOverride, called at test/integration/API/Teams/Feature.hs:94:44 in main:API.Teams.Feature

          Use -p '/increase to unlimited/' to rerun this test only.

I haven't looked at it, but it might be an unrelated flaky test. I'll re-run the CI to see if happens again.

[isovector]

potentially unpleasant question: would it be possible to skip this effect altogether, given we're planning to merge brig and galley soon, and it'll all just be function calls?

Sorry, just saw this. It's doable, but is currently blocking getting more polysemy into brig. Assuming CI and review passes, the work here is done. Up to you though!

[mdimjasevic]

There is another failure:

      delete team:                                                                     FAIL

        Exception: Assertions failed:

         1: 404 =/= 200

        

        Response was:

        

        Response {responseStatus = Status {statusCode = 200, statusMessage = "OK"}, responseVersion = HTTP/1.1, responseHeaders = [("Transfer-Encoding","chunked"),("Date","Fri, 07 Oct 2022 12:22:45 GMT"),("Server","Warp/3.3.17"),("Vary","Accept-Encoding"),("Content-Encoding","gzip"),("Content-Type","application/json;charset=utf-8")], responseBody = Just "{\"access\":[\"invite\"],\"access_role\":\"activated\",\"access_role_v2\":[\"team_member\",\"non_team_member\",\"service\"],\"creator\":\"23989a0d-4a23-4249-8975-87d0fa8e0831\",\"id\":\"88e5df0a-0a3d-460d-9686-2a7c863dbfdf\",\"last_event\":\"0.0\",\"last_event_time\":\"1970-01-01T00:00:00.000Z\",\"members\":{\"others\":[{\"conversation_role\":\"wire_admin\",\"id\":\"3dba3a8c-fe51-4ac6-a13b-dcda93c76d6a\",\"qualified_id\":{\"domain\":\"federation-test-helper.test-b5oqewkoxfih.svc.cluster.local\",\"id\":\"3dba3a8c-fe51-4ac6-a13b-dcda93c76d6a\"},\"service\":{\"id\":\"b5703907-b3f4-4cc8-9223-e58352f31faa\",\"provider\":\"c6c6bfc9-bb97-470c-bbd4-862c3f084215\"},\"status\":0},{\"conversation_role\":\"wire_admin\",\"id\":\"b3675ca5-3fdf-4a59-9385-2edbfb013051\",\"qualified_id\":{\"domain\":\"federation-test-helper.test-b5oqewkoxfih.svc.cluster.local\",\"id\":\"b3675ca5-3fdf-4a59-9385-2edbfb013051\"},\"status\":0}],\"self\":{\"conversation_role\":\"wire_admin\",\"hidden\":false,\"hidden_ref\":null,\"id\":\"23989a0d-4a23-4249-8975-87d0fa8e0831\",\"otr_archived\":false,\"otr_archived_ref\":null,\"otr_muted_ref\":null,\"otr_muted_status\":null,\"qualified_id\":{\"domain\":\"federation-test-helper.test-b5oqewkoxfih.svc.cluster.local\",\"id\":\"23989a0d-4a23-4249-8975-87d0fa8e0831\"},\"service\":null,\"status\":0,\"status_ref\":\"0.0\",\"status_time\":\"1970-01-01T00:00:00.000Z\"}},\"message_timer\":null,\"name\":null,\"protocol\":\"proteus\",\"qualified_id\":{\"domain\":\"federation-test-helper.test-b5oqewkoxfih.svc.cluster.local\",\"id\":\"88e5df0a-0a3d-460d-9686-2a7c863dbfdf\"},\"receipt_mode\":null,\"team\":\"115bf155-b5b7-4099-b5ff-473d1eb65b5a\",\"type\":0}", responseCookieJar = CJ {expose = []}, responseClose' = ResponseClose}

        CallStack (from HasCallStack):

          error, called at src/Bilge/Assert.hs:89:5 in bilge-0.22.0-inplace:Bilge.Assert

          <!!, called at src/Bilge/Assert.hs:107:19 in bilge-0.22.0-inplace:Bilge.Assert

          !!!, called at test/integration/API/Provider.hs:737:7 in main:API.Provider

        Use -p '(!/turn/&&!/user.auth.cookies.limit/)&&/delete team/' to rerun this test only.

[mdimjasevic]

@fisx , I find this a useful PR. Before we can merge Brig and Galley, we have to fully polysemise Brig, and this PR is bringing us in that direction.

[elland]

Is this correct?

[mdimjasevic]

There is another failure:

      delete team:                                                                     FAIL
> ```

I couldn't reproduce this once locally, and neither could @elland . I believe it's a flaky test. So re-running it once again in CI.