Use fake firehose client in react-storybook build

[islemaster]

Replaces firehose imports with a fake firehose client in the react-storybook build. This is a speculative fix for the storybook deploy failing on Circle, per Will's observation that the failures began with the addition of the AWS SDK in #14053.

This might not be a bad idea, even if it doesn't fix the Circle build on its own - we really shouldn't have storybook hitting AWS in any way.

[ewjordan]

Should we be concerned about the AWS SDK adding memory weight in the browser, as well?

If that's the case, we could switch over to the AWS Lambda version of this with a bit of work - we have functioning code there which imports Jquery instead of the AWS SDK, the unsolved bit is hooking it all up in Cloudformation.

[islemaster]

It would be interesting to check on whether Asher's PR changed the size of our bundles significantly, but I'd be surprised if it did - it wasn't the first change to add the aws sdk to our project, so we're already shipping that weight somewhere. In the case of this error, we're running out of memory during our build step, not at runtime, so I'm guessing it's a fairly independent issue.

[islemaster]

Update: This change didn't deploy successfully, we still had the out of memory error. It looks like we're also building source maps, which is not something we really need in the deployed version. I'll try and figure out how to turn that off.

[islemaster]

Wow... that didn't work either. In fact, it failed in the exact same place - creating chunk assets - and still has SourceMapGenerator_serializeMappings in the reported security context. What is going on?

We shoudn't be blocking staging on a failed storybook deploy, so I'm going to disable the storybook deploy altogether for now and troubleshoot tomorrow.

[islemaster]

I've reduced this PR to its first commit, which stubs firehose in storybook builds. This apparently resolved an issue for @Erin007 today, so I'd like to merge just this change, even though it doesn't resolve the Circle storybook deploy. Following up on a different branch with further work on re-enabling storybook deploy.

[ashercodeorg]

LGTM. Just so that I'm clear, the issue isn't that test is trying to send Firehose API calls (it shouldn't be, the method returns early on local and test environments). The issue is the memory overhead of loading the AWS SDK?

[islemaster]

I'm not even sure it's memory overhead at this point - I can't see all of the error @Erin007 posted from storybook, but it looks more like an initialization problem caused just by importing the sdk in an unusual context. But I do know we're never going to need that sdk in storybook and you guys have wrapped it nicely (thank you for that!) so we might as well stub it to remove any chance of it being the issue. 🤷‍♂️

[Erin007]

The error was basically that those node modules didn't exist or couldn't be found.

[islemaster]

Hmm... that might have been solved with just a yarn call to install missing dependencies.