Deprecate Stream Enrich and enrich-rabbitmq

[spenes]

No description provided.

[benjben]

Great work @spenes !

I wonder if we should merge common/ and common-fs2. On one hand it's probably good to keep common/ as is with the just the core logic of the adapters/enrichments. But on the other hand it sounds weird to have two modules with some common logic. Given that Scala Common Enrich is published as a library to Maven in case someone wants to use it in a custom app, I think that we should leave it as is.

[benjben]

I don't like that but changing this would require a lot of refactoring in the tests. Instead I would prefer us to make it possible to create an IgluCirceClient manually without having to parse a config file, so that we don't get a IO[ ].

[benjben]

Apart from removing Stream Enrich and enrich-rabbitmq, I took the opportunity to do some cleaning in order to make the code easier to understand.

• HttpClient is now passed explicity as a parameter to the places that need it, rather than being passed implicitly everywhere. This makes it easier to follow where it is initialized, and it is much more clear what uses the HTTP client. For instance a RemoteAdapter now has its own HTTP client defined in the class, rather than passing it implicity when we convert to a RawEvent.

• Now the only implicit s are the ones from Cats Effect and from outside Enrich codebase (e.g. RegistryLookup). For instance

def build[F[_]: Monad: CreateForex: CreateIabClient: CreateIpLookups: CreateOWM: CreateParser: CreateUaParserEnrichment: sqlquery.CreateSqlQueryEnrichment: apirequest.CreateApiRequestEnrichment]

became

def build[F[_]: Async: Clock: ContextShift]

• I did some cleaning in the package names, so that it's more clear where imports come from. For instance

package com.snowplowanalytics.snowplow.enrich.common
package adapters
package registry

became

package com.snowplowanalytics.snowplow.enrich.common.adapters.registry

• A cached thread pool is now used for the blocking calls of the unit tests

• I removed cats.Id entirely from the codebase. This led to many changes in the tests because now all the results are wrapped in the effect.

[spenes]

Looks good to me! However, Github doesn't allow me to approve the PR since I am the one opened the PR. But, from my point of view, it is good to go.