fix(ras-acc): handle recaptcha submit errors

[chickenn00dle]

All Submissions:

• [ ] Have you followed the Newspack Contributing guideline?
• [ ] Does your code follow the WordPress' coding standards and VIP Go coding standards?
• [ ] Have you checked to ensure there aren't other open Pull Requests for the same update/change?

Changes proposed in this Pull Request:

Closes https://app.asana.com/0/1207817176293825/1207878733150051/f

This PR addresses a problem where Firefox could end up with a strange recaptcha connection error and prevent v2 form submissions. This PR addresses this by adding an error callback and forcing a recaptcha refresh when an error is detected.

How to test the changes in this Pull Request:

Be sure to also checkout the following PRs before testing:

• https://github.com/Automattic/newspack-blocks/pull/1859

1. As logged out user, try to trigger the connection error (see below note)
2. Confirm a recaptcha error appears within the auth or checkout modal and not as an alert
3. Confirm resubmitting the form is successful

Note: This one is a bit tricky to reproduce. I found the best way to get the connection error to appear is to:

1. Use Firefox
2. Enable v2 recaptcha
3. Set recaptcha slider all the way down via Admin console: https://www.google.com/recaptcha/admin
4. Repeatedly go through the auth registration + checkout flow via a donation block, logout, go through the regular auth registration flow via signin button and repeat a handful of times

But truthfully not sure exactly what causes the issue...

Other information:

• [ ] Have you added an explanation of what your changes do and why you'd like us to include them?
• [ ] Have you written new tests for your changes, as applicable?
• [ ] Have you successfully ran tests with your changes locally?

[chickenn00dle]

I hope its okay @dkoo, but I've assigned you for review since you have the most context with recaptcha!

[dkoo]

@chickenn00dle I like how simple this fix ended up being. It doesn't necessarily avoid the issue, but it does create a workaround that seemingly works. And it feels like an edge case anyway so hopefully not many users will be affected.