Reset context on resurface

[Hakerh400]

Fixes #1292
Did a lot of refactoring.
If any of these is better to stay as it was implemented before, I can revert it:

• Hid some non-standard context prototype accessors
• Replaced deprecated __defineGetter__ with defineProperty
• Added new private method _updateStateInfo to track when the surface gets reset
• Hid lastFillStyle and similar properties
• Reduced some code duplication

---

• Updated the changelog
• Added a test

[zbjornson]

Really like the direction this is headed 👍. Hope to read it more closely tonight.

[zbjornson]

Looks great! Sorry for being slow to review. Just a few little bits of suggested cleanup, but otherwise good with me to land!

[Hakerh400]

Added some commits and rebased onto master. I'll add createImageData, but want to make sure everything so far is ok before adding it.

Checking if one v8::Value is an instance of another v8::Value doesn't seem trivial, unless I'm missing something. It can be done by constructing a new function using global.Function constructor that performs JS instanceof operator on the arguments and returns the result, but it is very slow.

Two questions: the BMP PR added a line into the changelog section for 2.1.0, but it landed after 2.1.0, right? I can move it in the unreleased section, if that is the case. Another question: tests registerFont and Context2d#measureText().width are flaky and fail randomly (see this, this, this, this (before restart), and this). Should we increase their timeout? If yes, let me know, all of that can be done in this PR.

[zbjornson]

the BMP PR added a line into the changelog section for 2.1.0, but it landed after 2.1.0, right?

Yeah, appears so. Thanks for finding.

tests registerFont and Context2d#measureText().width are flaky and fail randomly

I don't think increasing the timeout is the right thing to do (they shouldn't take more than 2000 ms, unless the host has some I/O issue with slow font loading?). I've only seen them be flaky in CI, not locally. Not sure what the answer is there. I guess we could try increasing the timeout in a branch and run CI a bunch of times.

[zbjornson]

Nice work!

[zbjornson]

Looks like this also fixes a bug in the original JS impl. 👍

[LinusU]

This looks great!! Thank you for the huge improvement 👏

I've left some small comments, but they don't have to block the merge, I can fix them later if you would prefer that 🎉

[LinusU]

Awesome 👏

[LinusU]

Thank you for this 🚀