Set 'Change >' and sublabel to selected embedded form rows

[yuki-stripe]

Summary

Adds logic to set and unset the "Change >" + sublabel.

Motivation

https://jira.corp.stripe.com/browse/MOBILESDK-3057

Testing

See new tests

Changelog

See changelog

[github-actions]

🚨 New dead code detected in this PR:

FormSpecProvider.swift:60 warning: Function 'load()' is unused

Please remove the dead code before merging.

If this is intentional, you can bypass this check by adding the label skip dead code check to this PR.

ℹ️ If this comment appears to be left in error, double check that the flagged code is actually used and/or make sure your branch is up-to-date with master.

[porter-stripe]

This doesn't seem to work with co-branded cards, when I enter a co-branded (Visa/CB) card and choose CB, the copy says "Visa ...". we need to also check the preferred network (if it exists) on the card params. Can we add a test for this too?

params.paymentMethodParams.card?.networks?.preferred

There's also a weird edge case where you can fill out a form for a co-branded card and don't select a brand since it's optional, so the card brand gets decided on the backend. Not exactly sure what we show there.

[yuki-stripe]

Thanks for catching this! Fixed it up. Are there things we can do to make it harder to make this mistake?

[porter-stripe]

We did something similar to what you did with func brand(for card: STPPaymentMethodCardParams?) -> STPCardBrand on STPPaymentMethodCard. But this is the first instance AFAIK where we have needed to read it off the STPPaymentMethodCardParams. Maybe we "internally" deprecate STPCardBrandUtilities.stringFrom(brand) somehow, since the card number is not the single source of truth for the card brand for a number.

[porter-stripe]

You can probably ignore the dead code flag, it's flagging it b/c that function is only used in tests. But maybe that function should be in a test file?

[yuki-stripe]

I approved a bunch of snapshot tests that are not quite right; the labels can get squished a bit w/ the hidden "Change" button. Debugging & fixing this is too toilsome with the current RowButton architecture relative to the tiny UI problem here, so I'm going to wait to fix this until we refactor RowButton as a GA fast-follow (likely after we submit for Implementation Review & finish the next major version) @porter-stripe