Bump Circle CI docker image

[FrederikBolding]

Explanation

Stops using deprecated circleci/node:16-browsers image and uses cimg/node:16.20-browsers instead.
https://circleci.com/developer/images/image/cimg/node

This updates the Node version we run in CI and fixes the following SES bug: nodejs/node#43496

A few other changes were made to accommodate the new Docker image:

• Stopped using xset to check for X-server, this would cause tests to timeout
• Added sudo apt-get update to the Chrome installation script because of missing packages required for Chrome installation
• Adjusted Firefox binary location to opt/firefox/firefox (we might not have been running the FF version we thought before)

Don't mind the commit log 😄

[metamaskbot]

Builds ready [12335a6]

• builds: chrome, firefox
• builds (beta): chrome
• builds (flask): chrome, firefox
• build viz: Build System
• mv3: Background Module Init Stats
• mv3: UI Init Stats
• mv3: Module Load Stats
• mv3: Bundle Size Stats
• mv2: E2e Actions Stats
• code coverage: Report
• storybook: Storybook
• typescript migration: Dashboard
• all artifacts
• bundle viz:

  • background: 0, 1, 2, 3, 4
  • common: 0, 1, 2, 3, 4
  • content-script: 0
  • contentscript: contentscript
  • disable-console: disable-console
  • inpage: inpage
  • policy-load: policy-load
  • sentry-install: sentry-install
  • ui: 0, 1, 10, 2, 3, 4, 5, 6, 7, 8, 9

Page Load Metrics (1580 ± 57 ms)

Platform	Page	Metric	Min (ms)	Max (ms)	Average (ms)	StandardDeviation (ms)	MarginOfError (ms)
Chrome	Home	firstPaint	94	190	116	21	10
		domContentLoaded	1380	1842	1552	114	55
		load	1398	1842	1580	120	57
		domInteractive	1380	1842	1552	114	55

Bundle size diffs

• background: 0 bytes
• ui: 0 bytes
• common: 0 bytes

[Gudahtt]

We had found that the e2e tests would intermittently fail unless this was running. This took a long time for us to debug. Eventually we settled on this solution after finding that the Chromium project uses a similar strategy for their own internal test suite.

I wonder why it's not required anymore... hopefully this won't bring the intermittent failures back.

[FrederikBolding]

I'm not sure.. None of the Chrome E2E tests would start when this was being run with it. It seemed like it would never resolve 🤔

Perhaps xset is not available on the new image? It's curious that it didn't throw though.

[Gudahtt]

Sorry I'm getting mixed up; this wasn't related to the e2e test failures I was thinking of. I went back to check the logs. I think we hypothesized that some recurring failures were due to this, but that was never found to be the case.

I used X server to get the e2e tests running on GitHub Actions in an experimental PR I was working on on-and-off for months. Not an intermittent failure issue, just basic compatibility. That's what I remember this being critical for. That PR was closed, no part of that landed in develop.

It looks like this script was added to make the tests run correctly locally on M1 macbooks. Same solution but a different problem than I was thinking of. I'll test this PR locally on a macbook.

[FrederikBolding]

It is only enabled in CI though 🤔

[Gudahtt]

Ah, interesting. That condition was not present in the original PR where this was introduced.

[FrederikBolding]

I would say we remove it for now and we can investigate further if it turns out to introduce intermittent test failures?

[Gudahtt]

Disabled locally here: #12043
Weird 🤷 I guess we can assume it's fine to omit it then.

[Gudahtt]

LGTM!