Fixing response clone test and ironing utils.js for different UInt8Array behavior

[youennf]

No description provided.

[hoppipolla-critic-bot]

Critic review: https://critic.hoppipolla.co.uk/r/6678

This is an external review system which you may optionally use for the code review of your pull request.

In order to help critic track your changes, please do not make in-place history rewrites (e.g. via git rebase -i or git commit --amend) when updating this pull request.

[wpt-pr-bot]

Reviewers for this pull request are: @jdm and @youennf.

[jdm]

It's not clear to me that papering over differences in the Uint8ArrayBuffer constructor is helpful. That's a standard API with well-defined semantics; if it's not implemented according to the specification, why do we want to hide that fact?

[youennf]

I believe all browsers accept "new Uint8Array()".
Some browsers seem to accept "new Uint8Array(undefined)" while others do not.
From a quick reading of the spec, the second behavior seems compliant.
Given that, since for empty streams, undefined may be passed to Uint8Array, I added a check in this method. I am fine changing the caller function though.

[youennf]

If needed, I can split the patch to at least fix the response clone test

[jdm]

Thinking about it further, I'm fine with this change. There are other tests that can focus on ES type specification behaviour.