Proposed changes

When using automationProtocol devtools sometimes the Pupeteer execution context is destroyed.
Currently, when that happens, we attempt to wait for a page loaded event to be emitted, after it is emitted, we retry the command. I recently discovered a case where the page load is never emitted, therefore the tests would hang for a long time and then finally error with

Error: page load timeout
    at Timeout._onTimeout (file:///opt/atlassian/pipelines/agent/build/e2e/node_modules/devtools/build/devtoolsdriver.js:141:73)
    at listOnTimeout (node:internal/timers:568:17)
    at processTimers (node:internal/timers:511:7)

This same use case also benefits from an immediate retry of the command.
When rerun (without first waiting for the page load that is never emitted) the command successfully executes and returns the desired result.

I've implemented a 3 retry limit for a command, therefore the previous behavior is maintained but only after attempting the command again a couple of times.

You can see a demo for this with the actual use case in question.

Nothing was mocked for this demo, a real test executes a command with a destroyed context and then recovers after rerunning the command. This same demo would hang eventually with Error: page load timeout if these changes were not implemented, as the page load is never completed.

Types of changes

• Bugfix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Documentation update

Checklist

• I have read the CONTRIBUTING doc
• I have added tests that prove my fix is effective or that my feature works
• [] I have added necessary documentation (if appropriate)
• I have added proper type definitions for new commands (if appropriate)

Further comments

Reviewers: @webdriverio/project-committers